Approximate optimal voltage control method and system for large-scale networked converter

Abstract

The invention provides an approximate optimal voltage control method and system for a large-scale networked converter. The method comprises the steps that a graph limit matrix is determined according to the electrical distance between converter access nodes and the information transmission efficiency between converters; constructing an active / reactive power-voltage approximate optimal control model of the converter based on the determined graph limit matrix, wherein the model comprises a reactive power-voltage control target function of the converter m and an active power-voltage control target function of the converter m based on a defined voltage function; and constructing a Nash fitness function according to the Nash equilibrium condition, and solving according to the Nash fitness function to determine an approximate optimal active / reactive power control strategy of the large-scale networked converter meeting the voltage safety constraint.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system operation and control, and in particular relates to a method and system for approximate optimal voltage control of a large-scale networked converter. Background Art

[0002] The statements in this section merely provide background information related to the present invention and do not necessarily constitute prior art.

[0003] In power systems with a high proportion of renewable energy, the grid structure changes, and renewable energy is connected to the grid through power electronic converters, which brings many challenges to the power system in terms of voltage safety. First of all, in the distribution network dominated by new energy, the converter is required to assume the responsibility of voltage regulation. In addition, with the large-scale access of converters to the grid, the power flow distribution of the system becomes a two-way flow. When the power injected into the grid by the converter at a certain grid connection point is too high, it will cause the voltage to exceed the limit, endangering the stable operation of the power system. Given that power changes have a great impact on voltage, and considering that the converter has a fast response speed and power regulation capability, controlling the output power of the converter is an effective means to solve the problem of voltage exceeding the limit at the converter grid connection point.

[0004] At present, there are three types of power control methods for converters: local, centralized, and distributed. Local control can be simply completed using the converter's own autonomous control function, which can quickly respond to local voltage, but the simple local strategy cannot be coordinated with other converters and lacks a global view. Centralized control is a unified command control by the central controller based on global information, which can achieve global optimization. However, when a large number of high-penetration converters are connected, the number of control variables increases greatly, and the centralized control process is complicated and cannot meet the response speed requirements.

[0005] Therefore, it is necessary to study the distributed control structure of the converter. The challenge of distributed control is mainly consensus. However, most of the existing research works have not reached a consensus on the reactive power and active power of each converter, and have not fully utilized the resources of each converter in the voltage regulation process. Summary of the invention

[0006] In order to overcome the above-mentioned deficiencies of the prior art, the present invention provides a method for approximate optimal voltage control of a large-scale networked converter, which can achieve reasonable and accurate control of the converter grid connection point voltage.

[0007] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0008] In a first aspect, a method for approximate optimal voltage control of a large-scale networked converter is disclosed, comprising:

[0009] Determine the graph limit matrix according to the electrical distance between the converter access nodes and the information transmission efficiency between the converters;

[0010] Based on the determined graph limit matrix, an active / reactive-voltage approximate optimal control model of the converter is constructed, and the model includes:

[0011] Setting the objective function of reactive power-voltage control of converter m and the objective function of active power-voltage control of converter m based on the defined voltage function;

[0012] A Nash fitness function is constructed according to the Nash equilibrium condition, and the Nash fitness function is solved to determine the approximate optimal active / reactive power control strategy for large-scale networked converters that meets voltage security constraints.

[0013] As a further technical solution, the process of obtaining the electrical distance between the converter access nodes is as follows:

[0014] The voltage sensitivity coefficient of the distribution network is derived based on the Jacobian matrix of the power system;

[0015] The electrical distance between nodes is determined based on the voltage sensitivity coefficient.

[0016] As a further technical solution, it also includes: obtaining the inverse matrix of the Jacobian matrix of the power system, determining the differential relationship between voltage and power through the inverse matrix of the Jacobian matrix of the power system, and constructing a voltage sensitivity index based on the inverse matrix of the Jacobian matrix to reflect the sensitivity of each node voltage to the power of each node.

[0017] As a further technical solution, the information transmission efficiency between the converters is obtained by:

[0018] It is obtained based on the shortest distance between the communication nodes corresponding to converter m and converter l, the delay occurring during information transmission, and the shortest distance between the communication nodes corresponding to converter m and converter e.

[0019] As a further technical solution, the graph limit matrix includes an active-graph limit matrix and a reactive-graph limit matrix;

[0020] Among them, w p,ml Indicates the active power intensity of converter l on converter m, W p,ml represents the degree of active power-voltage interaction between converter m and converter l, W p,ml ∈[0,1];

[0021] w q,ml Indicates the reactive power intensity of converter l on converter m; W q,ml represents the reactive power-voltage interaction degree of converter m and converter l, W q,kl∈[0,1].

[0022] As a further technical solution, the objective function of reactive power-voltage control of the converter m is used to reduce the voltage over-limit of the converter m while compensating the reactive power as little as possible.

[0023] As a further technical solution, the objective function of the active power-voltage control of the converter m is used to reduce the voltage over-limit of the converter m while reducing the active power as little as possible.

[0024] As a further technical solution, a Nash fitness function is constructed according to the Nash equilibrium condition, specifically:

[0025] Assume that the active decision variable of converter m is P m , all values ​​of the decision variables constitute the active strategy set of converter m, and the decision variables of all converters constitute the mean field. If there is a strategy combination under which the cost of each converter is the lowest under the opponent's strategy, then this strategy combination is called Nash equilibrium;

[0026] The key to solving Nash equilibrium using particle swarm optimization is to calculate the value of Nash fitness function and construct active-Nash fitness function to determine whether the graphon mean field games (GMFG) model reaches Nash equilibrium.

[0027] If and only if the strategy set reaches a Nash equilibrium, f pNash is equal to the minimum value 0. If the active power control strategy of any converter is not at the equilibrium point, f pNash >0, the Nash equilibrium is the global optimal point of the active-Nash fitness function.

[0028] As a further technical solution, the approximate optimal active / reactive power control strategy for large-scale networked converters that meets voltage security constraints is determined as follows:

[0029] Step S042: Determine reactive power margin, assuming that the converter m in the partition has put into full adjustable reactive capacity, determine whether the voltage exceeds the limit, if the voltage exceeds the limit, proceed to step S043; if the voltage does not exceed the limit, proceed to step S045;

[0030] Step S043: Active voltage regulation calculation, firstly, all adjustable reactive capacity is put into use, and on this basis, the active Nash equilibrium point of each converter in the partition is solved by using the improved particle swarm algorithm;

[0031] Step S044: according to the active voltage control result, the converter m updates the reactive power compensation power limit and re-determines whether the voltage exceeds the limit. If it exceeds the limit, the process proceeds to step S045; if it does not exceed the limit, the voltage control is completed;

[0032] Step S045: reactive voltage regulation calculation: using an improved particle swarm algorithm to solve the reactive Nash equilibrium point of each converter in the partition;

[0033] Step S046: Based on the reactive voltage control result, determine whether the voltage exceeds the limit. If so, proceed to step S043; if not, the voltage control is completed.

[0034] In a second aspect, a large-scale networked converter approximate optimal voltage control system is disclosed, comprising:

[0035] The acquisition module is configured to: determine a graph limit matrix according to an electrical distance between converter access nodes and an information transmission efficiency between converters;

[0036] The modeling module is configured to: construct an active / reactive-voltage approximate optimal control model of the converter based on the determined graph limit matrix, the model comprising:

[0037] Setting the objective function of reactive power-voltage control of converter m and the objective function of active power-voltage control of converter m based on the defined voltage function;

[0038] The optimization module is configured to: construct a Nash fitness function according to the Nash equilibrium condition, and solve the Nash fitness function to determine an approximately optimal active / reactive power control strategy for a large-scale networked converter that meets the voltage security constraint.

[0039] One or more of the above technical solutions have the following beneficial effects:

[0040] The large-scale networked converter system graph limit mean field game model constructed by the technical solution of the present invention takes into account both the physical topology and the communication structure, and adopts a dual-scale metric based on the improved electrical distance between nodes and information transmission efficiency to construct a graph limit matrix to approximately characterize the degree of interaction between converters, thereby forming a mean field function, and constructing a game function of reactive power-voltage control to reduce reactive power compensation of converters and active power-voltage control to avoid active power reduction according to the mean field function, thereby more accurately representing the mean field effect.

[0041] The technical solution of the present invention constructs a Nash fitness function according to the Nash equilibrium equation, and solves the graph limit mean field game model through an improved particle swarm algorithm, which has the advantages of fast convergence and preventing the algorithm from falling into local optimum.

[0042] The technical solution of the present invention can obtain an approximate optimal active / reactive power control strategy for the converter that meets the voltage safety constraints based on the solution results of the graph limit mean field game model, thereby achieving effective control of the voltage with less power changes, and the control is accurate and efficient.

[0043] In the technical solution of the present invention, each converter reaches a consensus through the objective function of the graph limit mean field game (each converter confirms how to output to improve the voltage through GMFG). Utilizing the resources of the converter means utilizing the active power and reactive power of all converters with voltage regulation capability.

[0044] Advantages of additional aspects of the present invention will be given in part in the following description, and in part will become obvious from the following description, or will be learned through practice of the present invention. BRIEF DESCRIPTION OF THE DRAWINGS

[0045] The accompanying drawings in the specification, which constitute a part of the present invention, are used to provide a further understanding of the present invention. The exemplary embodiments of the present invention and their descriptions are used to explain the present invention and do not constitute improper limitations on the present invention.

[0046] Figure 1 is a flow chart of the approximate optimal voltage control strategy for a large-scale networked converter in the first embodiment of the present invention;

[0047] Figure 2 It is a limit mean field game control block diagram of the converter additional diagram in the first embodiment of the present invention;

[0048] Figure 3 is a calculation flow chart of the improved particle swarm algorithm in the first embodiment of the present invention;

[0049] Figure 4 Schematic diagram of the topological structure and partitioning of the IEEE 118-node system in Embodiment 1 of the present invention;

[0050] Figure 5 is a schematic diagram of a converter communication network in Embodiment 1 of the present invention;

[0051] Figure 6 is a graph limit matrix result diagram in the first embodiment of the present invention, wherein (a) is an active-graph limit matrix, and (b) is a reactive-graph limit matrix;

[0052] Figure 7 It is the voltage distribution before and after the extreme mean field game control in the first embodiment of the present invention;

[0053] Figure 8 It is the power change of each converter in the partition 6 before and after the extreme mean field game control in the first embodiment of the present invention, where (a) is the active power reduction amount, and (b) is the reactive power absorption amount;

[0054] Fig. 9 : It is the power change of each converter in the partition 1 and the partition 2 before and after the extreme mean field game control in the first embodiment of the present invention, wherein (a) is the reactive power absorbed by each converter in the partition 1 and (b) is the reactive power absorbed by each converter in the partition 2;

[0055] Fig.10 It is a structural block diagram of the approximate optimal voltage control system of a large-scale networked converter in the second embodiment of the present invention. DETAILED DESCRIPTION

[0056] It should be noted that the following detailed descriptions are exemplary and are intended to provide further explanation of the present invention. Unless otherwise specified, all technical and scientific terms used herein have the same meanings as those commonly understood by those skilled in the art to which the present invention belongs.

[0057] It should be noted that the terms used herein are for describing specific embodiments only and are not intended to be limiting of exemplary embodiments according to the present invention.

[0058] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0059] Embodiment 1

[0060] like Figure 1 As shown, this embodiment discloses a method for approximate optimal voltage control of a large-scale networked converter, including:

[0061] Step S01: obtaining relevant parameters of the power network and the communication network, and constructing a graph limit matrix of a large-scale networked converter based on the relevant parameters of the power network and the communication network;

[0062] In this step, the acquired parameters related to the power network and the communication network include at least the load, branch resistance, branch reactance of the distribution network and the delay of the communication network, which are used to subsequently construct the graph limit matrix.

[0063] When using graphical limits to characterize the degree of interaction between converters, it includes:

[0064] The voltage sensitivity coefficient of the distribution network is derived based on the Jacobian matrix J;

[0065] Determining an improved inter-node electrical distance based on a voltage sensitivity coefficient;

[0066] Characterize the efficiency of information transmission based on the shortest communication path;

[0067] Characterize graph limits using improved dual-scale metrics of inter-node electrical distance and information transmission efficiency;

[0068] According to the influence mechanism of the converter on the node voltage in the distribution network, an active-graph limit matrix and a reactive-graph limit matrix are formed respectively;

[0069] Step S02: constructing an active / reactive-voltage approximate optimal control model of the converter based on the graph limit mean field game theory;

[0070] When constructing a graph limit mean field game model for a large-scale networked converter system, it includes: determining a suitable voltage function according to the characteristics of the voltage control problem; forming a mean field function according to the graph limit matrix, and constructing the game functions of reactive power-voltage control and active power-voltage control of the graph limit mean field game model based on the voltage function;

[0071] Step S03: Using the Nash fitness function constructed based on the Nash equilibrium principle as the fitness function of the improved particle swarm algorithm to solve the graph limit mean field game GMFG model;

[0072] When using the improved particle swarm algorithm to solve the graph limit mean field game model, the Nash fitness function is constructed based on the Nash equilibrium principle, and the quality of the particle swarm algorithm solution is evaluated by the Nash fitness function. If and only if the algorithm solution reaches Nash equilibrium, the Nash fitness function is equal to the minimum value 0, and the algorithm solution is globally optimal.

[0073] Step S04: Propose a "reactive power first, active power later" large-scale networked converter coordinated voltage control strategy, thereby determining a near-optimal active / reactive power control strategy for large-scale networked converters that meets voltage safety constraints.

[0074] In this step, partition control is adopted. By reasonably partitioning the distribution network, the coupling between converters is distinguished, so that converters with strong coupling in the same partition can communicate with each other to achieve coordinated voltage control, reducing the communication burden and control complexity.

[0075] Based on the voltage control method of the converter, the above-mentioned converter coordinated voltage control strategy of "reactive power first, active power later" is proposed by judging whether the reactive power compensation margin of the converter is sufficient, so as to determine the approximate optimal active / reactive power control strategy of large-scale networked converters that meets the voltage safety constraints. Specifically:

[0076] When the reactive power compensation margin of the converter is sufficient, the reactive power is directly adjusted for voltage control; when the reactive power compensation margin of the converter is insufficient, all reactive power compensation capacity is put into use, and then the active power output by the converter is reduced for voltage control, and the reactive power compensation margin is further increased.

[0077] In this implementation example, it should be noted that the graph limit can be regarded as a real-valued function defined on the unit square, represented by a bounded symmetric measurable function W:[0,1]2→[0,1], which can be interpreted as a weighted graph on the node set [0,1]. The graph limit can represent the relationship strength or connection probability between any two nodes in an infinite graph. When applying the graph limit method to solve power system problems, the interaction function between intelligent agents can be established according to actual conditions. The interaction strength between converters is affected by the topological connection and communication conditions of the power grid.

[0078] In step S01, a dual-scale metric based on improved inter-node electrical distance and information transmission efficiency is used to characterize the graph limit; specifically,

[0079] Step S011: deriving the voltage sensitivity coefficient of the distribution network according to the Jacobian matrix J. The Jacobian matrix is ​​obtained by calculating the derivative of the power flow equation;

[0080] According to the Jacobian matrix J of the power system, when the power of the distribution network node changes, it will affect the voltage amplitude and phase angle of the node.

[0081]

[0082] Where ΔP and ΔQ represent the changes of active power and reactive power injected into the node, respectively; Δδ and ΔU represent the changes of node voltage phase angle and amplitude, respectively.

[0083] By the inverse matrix J of the power flow Jacobian matrix -1 The differential relationship between voltage and power can be determined.

[0084]

[0085] According to J -1 The voltage sensitivity index is constructed to reflect the sensitivity of each node voltage to the power of each node.

[0086]

[0087] In the formula, and The degree of change in the voltage amplitude at node i when a unit amount of active and reactive power is injected into node j. and They represent the voltage-active power sensitivity coefficient and the voltage-reactive power sensitivity coefficient, respectively, reflecting the sensitivity of the voltage amplitude of node i to the unit amount of active power and reactive power injected into node j.

[0088] Step S012: determining an improved inter-node electrical distance according to the voltage sensitivity coefficient.

[0089] In the traditional definition of electrical distance, only the location of the node in the distribution network is considered. After the converter is connected, the impact of power changes caused by the connection of other nodes to the converter needs to be considered. Therefore, considering the location of the nodes in the distribution network and the impact of the converter connection, based on the voltage sensitivity coefficient, the improved electrical distance d between nodes i and j is defined ij .

[0090]

[0091] After the converter is connected to the grid, it will be indirectly affected by the other node converters based on the topology of the grid. Therefore, the electrical distance between the converter access nodes can form a part of the graph limit matrix to characterize the interaction strength between converters.

[0092] Step S013: Characterize the information transmission efficiency based on the shortest communication path.

[0093] The transmission efficiency of information is affected by packet loss, delay, etc. In addition, in the communication network, in order to reduce costs, the data packet needs to choose the shortest path when transmitting between information nodes. Therefore, the information transmission efficiency is described based on the shortest path, as follows:

[0094]

[0095] In the formula, w 2,ml represents the information transmission efficiency between converter m and converter l; R ml represents the shortest distance between the communication nodes corresponding to converter m and converter l; M represents the communication node group corresponding to all converters in the partition where converter m is located; T c Indicates the delay that occurs when information is transmitted.

[0096] Step S014: Determine the graph limit matrix according to the improved inter-node electrical distance and information transmission efficiency.

[0097] In medium and low voltage distribution networks, active power mainly affects voltage through resistance, and reactive power mainly affects voltage through reactance. Therefore, the active-reactive limit matrix W is defined respectively. p and reactive-graph limit matrix W q , where W p The elements and W q The elements are:

[0098]

[0099] In the formula, w p,ml represents the active action intensity of converter l on converter m; considering that the limit of the figure is a symmetric matrix, it is processed into W p,ml , W p,mlrepresents the degree of active power-voltage interaction between converter m and converter l, W p,ml ∈[0,1]. q,ml Indicates the reactive power intensity of converter l on converter m; W q,ml represents the reactive power-voltage interaction degree of converter m and converter l, W q,kl ∈[0,1]. 1 and λ 2 is the weight parameter, λ 1 +λ 2 =1.

[0100] In formula (8), W p,ml is the calculation method of each element in the matrix Wp; W in formula (9) q,ml It is the calculation method of each element in the matrix Wq.

[0101] W p,ml and W q,ml The larger the value, the greater the effect of converter 1 on converter m. It can also be understood that in voltage control, when the output power is the same, the greater the influence of converter 1 on the grid-connected voltage of converter m, and thus the greater the indirect influence on the output power of converter m.

[0102] As one or more implementation modes, in step S02, an active / reactive-voltage approximate optimal control model of the converter is constructed based on the GMFG theory. The GMFG additional control block diagram of the converter is as follows: Figure 2 As shown; specifically,

[0103] Step S021: Determine a suitable voltage function according to the characteristics of the voltage control problem.

[0104] The voltage control problem has the following characteristics: 1) The voltage control problem is a combination of local and global problems, that is, the voltage of each converter is affected by the active power and reactive power of all other converters, but the influence gradually weakens with the increase of electrical distance and communication distance. 2) The voltage control problem is a constrained optimization problem, where the constraint is the voltage threshold and the target is the total power loss, but there is no clear relationship between the constraint and the control. Define the voltage reference value u ref =1.00pu, the voltage needs to be controlled within the range of 0.95pu-1.05pu. The voltage function has different forms of expression. In order to promote the use of appropriate power to meet the requirements of voltage safety, the expression of the voltage function is defined as follows:

[0105]

[0106] In the formula, u i is the voltage at the grid connection point i of converter m; b 1 , b2 , b 3 and b 4 is the penalty coefficient. The specific meaning of the above formula is that the penalty is very small within the voltage safety range, and the penalty increases rapidly outside the safety range.

[0107] Step S022: Establishing an approximate optimal control model of active / reactive power-voltage of a large-scale networked converter based on the GMFG theory according to the voltage control method of the converter.

[0108] The goal of each agent in GMFG is to select a control policy o(x)∈O to maximize its benefit J x , where the mean field is expressed as:

[0109]

[0110] Based on the GMFG theory, an approximate optimal control model of active / reactive-voltage of large-scale networked converters is established. The control objectives include: 1) ensuring that the voltage at the converter grid connection point does not exceed the limit, giving priority to ensuring the safety and reliability of the system; 2) maximizing the active power output of the converter and improving the converter utilization; 3) minimizing the reactive power output of the converter and reducing line network losses.

[0111] When the reactive capacity margin of converter m is sufficient, the reactive power is directly adjusted for voltage control. The objective function of reactive power-voltage control of converter m is set as:

[0112]

[0113] Where Q(y) represents the reactive power output by converter k, |Q(y)| max (y), Q max (y) represents the reactive power compensation limit of converter k; γ q Represents the reactive power penalty coefficient. The above formula means that the converter m can mitigate the voltage over-limit while compensating the reactive power as little as possible.

[0114] Theoretical adjustable reactive power limit value Q of converter m at time t m,t and capacity S m The relationship is:

[0115]

[0116] From (13), it can be seen that when active power reduction is used for voltage regulation, specifically: when the reactive capacity margin of converter m is insufficient, reducing the active power output of the converter can increase the reactive power limit value while reducing the grid voltage, but the active power reduction should not be too large. The objective function of active power-voltage control of converter m is set as: ​

[0117]

[0118] Where y = l / M h , M h Represents partition C h The number of internal converters; P(y) represents the active power output by converter l, P(y)>P min (y), P min (y) represents the minimum value of active power allowed to be output by converter 1; γ p Represents the active power penalty coefficient. The above formula means that the converter m can mitigate the voltage over-limit while reducing the active power as little as possible.

[0119] The calculation of the Nash equilibrium of the GMFG model is an NP-hard problem. As one or more implementation methods, in step S03, the Nash fitness function constructed based on the Nash equilibrium principle is used as the fitness function of the improved particle swarm algorithm to solve the GMFG model, accelerate the convergence of the model and determine the global optimal solution; the calculation process of active power-voltage control is used for explanation, and the calculation process of reactive power-voltage control is similar, specifically:

[0120] Step S031: constructing a Nash fitness function according to the Nash equilibrium condition.

[0121] Assume that the active decision variable of converter m is P m , all the values ​​of the decision variables constitute the active strategy set O of converter m m = {P m |P m ∈[P m,min ,P m,max ]}. The cost of all converters is {J P,1 ,···,J P,m ,···,J P,Mh}, the decision variables of all converters constitute the mean field If there is a strategy combination Under this strategy, the cost of each converter is the lowest under the opponent's strategy. This strategy combination is called a Nash equilibrium, which satisfies the following conditions:

[0122]

[0123] The above formula is also equivalent to:

[0124]

[0125] The key to solving Nash equilibrium using particle swarm optimization is to calculate the value of Nash fitness function. According to formula (15), the active-Nash fitness function f is constructed. pNash , to determine whether the GMFG model reaches Nash equilibrium:

[0126]

[0127] Obviously, according to the definition of Nash equilibrium, f pNash is equal to the minimum value 0. If any active power control strategy of the converter is not at the equilibrium point, f pNash > 0. The Nash equilibrium is the global optimal point of the active-Nash fitness function.

[0128] Step S032: using an improved particle swarm algorithm to solve the objective function according to the Nash fitness function to obtain a power control strategy for the converter.

[0129] The improved particle swarm algorithm is used to solve the active power control strategy of the converter, and the active-Nash fitness function is used to evaluate the quality of the solution. The solution process is as follows: Figure 3 As shown. Each particle i has two attributes X representing its position and velocity respectively. i ,Vi. Among them, X i = {P 1 ,···,P m ,···,P Mh}, represents the active control strategy of each converter. Considering the constraints of the control strategy, in order to ensure that the particles are always within the feasible solution constraint range, the original particle swarm algorithm is improved. The speed and position update formula of particle i can be expressed as:

[0130]

[0131]

[0132]

[0133] In the formula, k is the number of iterations, k max is the maximum number of iterations; represents the optimal position of particle i in the first k iterations; represents the optimal position of all particles in the first k iterations. In other words, represents the global optimal solution of the first k iterations; ω k is the speed weight factor; ω min ,ω max 、c 1 and c 2 is a constant; rand is a random number uniformly distributed on [0,1].

[0134] The above expression is part of the solution algorithm to obtain the optimal active / reactive power control strategy. i = {P 1 ,···,Pm ,···,P Mh}, indicating the active / reactive power control strategy of each converter. Figure 3 As shown, according to each X i Solve f with the same parameters pNash Judge X i Is it the optimal control strategy?

[0135] In step S04, according to the voltage control mode of the converter, a converter coordinated voltage control strategy of "reactive power first, active power later" is proposed to determine the approximate optimal active / reactive power control strategy of the large-scale networked converter that meets the voltage safety constraint; specifically,

[0136] Step S041: Reasonably partition the distribution network. When a large-scale converter is connected to the grid, the converter can communicate and coordinate control only with the converters in the partition where it is located, avoiding problems such as slow response speed caused by too long communication links, reducing the communication burden, and partition control also reduces control complexity.

[0137] Step S042: Determine reactive power margin. Assume that partition C h The internal converter m puts all the adjustable reactive capacity into use to determine whether the voltage exceeds the limit. If the voltage exceeds the limit, the reactive power margin is insufficient, and the process proceeds to step S043; if the voltage does not exceed the limit, the process proceeds to step S045.

[0138] Step S043: Active voltage regulation calculation. First, all adjustable reactive capacity is put into use, and on this basis, the improved particle swarm algorithm is used to calculate C h The active Nash equilibrium point of each converter is solved to obtain the X when the active fitness function is 0. i .

[0139] Step S044: According to the active voltage control result, the converter m updates the reactive power compensation power limit Q m,max . And re-determine whether the voltage exceeds the limit. If it exceeds the limit, go to step S045; if it does not exceed the limit, the voltage control is completed.

[0140] Step S045: reactive voltage regulation calculation. Using the improved particle swarm algorithm to calculate C h The reactive Nash equilibrium point of each converter is solved to obtain the X when the reactive fitness function is 0. i .

[0141] Step S046: According to the reactive voltage control result, determine whether the voltage exceeds the limit. If it exceeds the limit, go to step S043; if it does not exceed the limit, the voltage control is completed.

[0142] The above steps assume that the partition C hThis is done if there are enough voltage control resources in the partition C. h After the active power of all converters in the zone is reduced to the limit value and the reactive compensation power reaches the limit value, if the voltage at the grid connection point of converter m still exceeds the limit, it is necessary to coordinate with converters in other zones to control the voltage. The rules are as follows:

[0143] For converter m at the voltage-limit node, remove partition C from the mth row of the priority and reactive-map limit matrix h The converters represented by the maximum value after the column where the inner converter is located communicate and cooperate to complete voltage control.

[0144] The exemplary technical solution of this embodiment constructs a limit mean field game model of a large-scale networked converter system diagram taking into account the physical topology and communication structure, and on this basis proposes a "reactive first, active later" converter coordinated voltage control strategy to achieve reasonable and accurate control of the converter grid-connected point voltage.

[0145] The graph limit mean field game model of a large-scale networked converter system constructed by the sub-technical solution of this embodiment takes into account both the physical topology and the communication structure, and uses the improved dual-scale metric of the electrical distance between nodes and the information transmission efficiency to characterize the graph limit and then construct the mean field function. Based on the mean field function, the game functions of reactive power-voltage control for reducing reactive power compensation of the converter and active power-voltage control for avoiding active power reduction are constructed.

[0146] In the process of solving the extreme mean field game model of the constructed large-scale networked converter system diagram, the sub-technical scheme of this embodiment adopts an improved particle swarm algorithm to solve the upper model, and uses the Nash fitness function constructed based on the Nash equilibrium principle as the fitness function of the improved particle swarm algorithm to obtain the approximate optimal active / reactive power control strategy for the large-scale converter system that meets the voltage safety constraints.

[0147] The sub-technical solution of this embodiment adopts partition control, and the converter can communicate and coordinate control only with the converters in the partition, reducing the communication burden and control complexity. And according to the voltage control mode of the converter, by judging whether the reactive compensation margin of the converter is sufficient, a "reactive first, active later" converter coordinated voltage control strategy is proposed.

[0148] This embodiment takes the improved IEEE 118-node calculation system as an example to further illustrate the specific implementation process of the present invention. Figure 4As shown. The voltage of the first node is set to 1.05pu, the total capacity of the 35 converters is 29.1MVA, and the installation position and installation capacity of each converter are shown in Table 1. In addition, the upper limit of the active power reduction of the converters at nodes 50 and 111 is set to 10% of the capacity, and the upper limit of the active power reduction of the converters at the remaining nodes is set to 15% of the capacity. The communication network of the converter is shown in Figure 5 As shown, in GMFG control, only converters located in the same partition exchange control information.

[0149] Table 1 Installation location and installation capacity of converter

[0150]

[0151] The active-diagram limit matrix and reactive-diagram limit matrix are as follows Figure 6 As shown in (a) and (b) in . When the improved electrical distance between the two converters is small enough or the communication distance is close enough, the matrix elements in the corresponding graph limit will approach 1. In addition, the minimum value of the active-graph limit matrix is ​​0.075, which is the degree of active-voltage interaction between the converter at node 31 and the converter at node 107. This means that when the voltage at node 31 exceeds the limit, the active power cut by the converter at node 107 has little effect on the voltage at node 31. The converters at the two nodes should avoid active coordinated control unless necessary to avoid waste of active power.

[0152] The improved particle swarm algorithm is parameterized, where the speed factor ω min and ω max are 0.4 and 0.9 respectively, and the coefficient c 1 and c 2 Take 1.8, the number of particles is 100, and the maximum number of iterations k max is 200. According to the limit matrix of the graph, the approximate optimal voltage control model of large-scale networked converters is solved. The grid connection point voltage of each converter before and after GMFG control is adopted is shown in Figure 7 After GMFG control, the grid-connected voltage of each converter is controlled within 1.050pu, with the highest voltage at 1.050pu and the lowest voltage at 0.987pu.

[0153] Only in Zone 1, Zone 2 and Zone 6, the voltage exceeds the limit. The converters in these areas are coordinated for voltage control based on the GMFG theory. The power changes of each converter in Zone 6 are shown in Figure 2. Figure 8As shown, (a) is a schematic diagram of active power reduction, and (b) is a schematic diagram of reactive power absorption. The total reactive power absorbed to complete voltage control is 0.722Mvar, and the total active power reduction is 0.371MW. Among them, the reactive power utilization rate of the converter at node 118 is as high as 89.67%, which indirectly proves that GMFG control does not cause excessive active power reduction. The converters in partitions 1 and 2 can solve the voltage over-limit problem only by reactive power compensation. The reactive power changes of each converter are shown in Fig. 9 As shown in Figure 1, the total absorbed reactive power of zone 1 is 0.225 Mvar, and the total absorbed reactive power of zone 2 is 0.197 Mvar.

[0154] Embodiment 2

[0155] The purpose of this embodiment is to provide a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the program.

[0156] Embodiment 3

[0157] The purpose of this embodiment is to provide a computer-readable storage medium.

[0158] A computer-readable storage medium stores a computer program, which executes the steps of the above method when executed by a processor.

[0159] Embodiment 4

[0160] The purpose of this embodiment is to provide a large-scale networked converter approximate optimal voltage control system, including:

[0161] The acquisition module is configured to: determine a graph limit matrix according to an electrical distance between converter access nodes and an information transmission efficiency between converters;

[0162] The modeling module is configured to: construct an active / reactive-voltage approximate optimal control model of the converter based on the determined graph limit matrix, the model comprising:

[0163] Setting the objective function of reactive power-voltage control of converter m and the objective function of active power-voltage control of converter m based on the defined voltage function;

[0164] The optimization module is configured to: construct a Nash fitness function according to the Nash equilibrium condition, and solve the Nash fitness function to determine an approximately optimal active / reactive power control strategy for a large-scale networked converter that meets the voltage security constraint.

[0165] Embodiment 5

[0166] The purpose of this embodiment is to provide a computer program product containing instructions, which, when running on a computer, enables the computer to execute the methods and functions involved in any of the above embodiments.

[0167] The steps involved in the apparatus of the above embodiment correspond to the method embodiment 1, and the specific implementation method can refer to the relevant description part of embodiment 1. The term "computer-readable storage medium" should be understood as a single medium or multiple media including one or more instruction sets; it should also be understood to include any medium that can store, encode or carry an instruction set for execution by a processor and enable the processor to execute any method in the present invention.

[0168] Those skilled in the art should understand that the modules or steps of the present invention described above can be implemented by a general-purpose computer device, or alternatively, they can be implemented by a program code executable by a computing device, so that they can be stored in a storage device and executed by the computing device, or they can be made into individual integrated circuit modules, or multiple modules or steps therein can be made into a single integrated circuit module for implementation. The present invention is not limited to any specific combination of hardware and software.

[0169] Although the above describes the specific implementation mode of the present invention in conjunction with the accompanying drawings, it is not intended to limit the scope of protection of the present invention. Those skilled in the art should understand that various modifications or variations that can be made by those skilled in the art on the basis of the technical solution of the present invention without creative work are still within the scope of protection of the present invention.

Claims

1. A method for approximate optimal voltage control of a large-scale networked converter, characterized in that: include: Determine the graph limit matrix according to the electrical distance between the converter access nodes and the information transmission efficiency between the converters; Based on the determined graph limit matrix, an active / reactive-voltage approximate optimal control model of the converter is constructed, and the model includes: Setting the objective function of reactive power-voltage control of converter m and the objective function of active power-voltage control of converter m based on the defined voltage function; A Nash fitness function is constructed according to the Nash equilibrium condition, and the Nash fitness function is solved to determine the approximate optimal active / reactive power control strategy for large-scale networked converters that meets voltage security constraints.

2. A large-scale networked converter approximate optimal voltage control method as claimed in claim 1, characterized in that: The process of obtaining the electrical distance between the converter access nodes is as follows: The voltage sensitivity coefficient of the distribution network is derived based on the Jacobian matrix of the power system; Determine the electrical distance between nodes based on the voltage sensitivity coefficient; It also includes: obtaining the inverse matrix of the Jacobian matrix of the power system, determining the differential relationship between voltage and power through the inverse matrix of the Jacobian matrix of the power system, and constructing a voltage sensitivity index according to the inverse matrix of the Jacobian matrix to reflect the sensitivity of each node voltage to the power of each node.

3. A large-scale networked converter approximate optimal voltage control method as claimed in claim 1, characterized in that: The information transmission efficiency between the converters is obtained by: It is obtained based on the shortest distance between the communication nodes corresponding to converter m and converter l, the delay occurring during information transmission, and the shortest distance between the communication nodes corresponding to converter m and converter e.

4. A large-scale networked converter approximate optimal voltage control method as claimed in claim 1, characterized in that: The graph limit matrix includes an active-graph limit matrix and a reactive-graph limit matrix; Among them, w p,ml Indicates the active power intensity of converter l on converter m, W p,ml represents the degree of active power-voltage interaction between converter m and converter l, W p,ml ∈[0,1]; w q,ml Indicates the reactive power intensity of converter l on converter m; W q,ml represents the reactive power-voltage interaction degree of converter m and converter l, W q,kl ∈[0,1].

5. The method for approximate optimal voltage control of a large-scale networked converter according to claim 1, characterized in that: The objective function of reactive power-voltage control of converter m is used to: reduce the voltage limit of converter m while compensating reactive power as little as possible; The objective function of the active power-voltage control of the converter m is to reduce the voltage over-limit of the converter m while reducing the active power as little as possible.

6. A large-scale networked converter approximate optimal voltage control method as claimed in claim 1, characterized in that: The Nash fitness function is constructed according to the Nash equilibrium condition, specifically: Assume that the active decision variable of converter m is P m , all values ​​of the decision variables constitute the active strategy set of converter m, and the decision variables of all converters constitute the mean field. If there is a strategy combination under which the cost of each converter is the lowest under the opponent's strategy, then this strategy combination is called a Nash equilibrium; The key to solving Nash equilibrium using particle swarm optimization is to calculate the value of Nash fitness function and construct active-Nash fitness function to determine whether the GMFG model reaches Nash equilibrium. If and only if the strategy set reaches a Nash equilibrium, f pNash is equal to the minimum value 0. If the active power control strategy of any converter is not at the equilibrium point, f pNash >0, the Nash equilibrium is the global optimal point of the active-Nash fitness function.

7. A large-scale networked converter approximate optimal voltage control method as claimed in claim 1, characterized in that: To determine the approximate optimal active / reactive power control strategy for large-scale networked converters that meets voltage security constraints, specifically: Step S042: Determine reactive power margin, assuming that the converter m in the partition has put into full adjustable reactive capacity, determine whether the voltage exceeds the limit, if the voltage exceeds the limit, proceed to step S043; if the voltage does not exceed the limit, proceed to step S045; Step S043: Active voltage regulation calculation, firstly, all adjustable reactive capacity is put into use, and on this basis, the active Nash equilibrium point of each converter in the partition is solved by using the improved particle swarm algorithm; Step S044: according to the active voltage control result, the converter m updates the reactive power compensation power limit and re-determines whether the voltage exceeds the limit. If it exceeds the limit, the process proceeds to step S045; if it does not exceed the limit, the voltage control is completed; Step S045: reactive voltage regulation calculation: using an improved particle swarm algorithm to solve the reactive Nash equilibrium point of each converter in the partition; Step S046: Based on the reactive voltage control result, determine whether the voltage exceeds the limit. If so, proceed to step S043; if not, the voltage control is completed.

8. A large-scale networked converter approximate optimal voltage control system, characterized in that: include: The acquisition module is configured to: determine a graph limit matrix according to an electrical distance between converter access nodes and an information transmission efficiency between converters; The modeling module is configured to: construct an active / reactive-voltage approximate optimal control model of the converter based on the determined graph limit matrix, the model comprising: Setting the objective function of reactive power-voltage control of converter m and the objective function of active power-voltage control of converter m based on the defined voltage function; The optimization module is configured to: construct a Nash fitness function according to the Nash equilibrium condition, and solve the Nash fitness function to determine an approximately optimal active / reactive power control strategy for a large-scale networked converter that meets the voltage security constraint.

9. A computer device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps of the method described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the program is executed by a processor, the steps of the method described in any one of claims 1 to 7 are performed.

Images