Power distribution network dynamic topology reconstruction method and system with high permeability photovoltaic power generation

Abstract

The invention belongs to the technical field of power system operation optimization, and particularly relates to a power distribution network dynamic topology reconstruction method and system with high permeability photovoltaic power generation, and the method comprises the steps: obtaining a power distribution network topology structure, network parameters, and node loads and photovoltaic predicted values; constructing a Markov decision process tetrad corresponding to the dynamic topology reconstruction problem; a sequential power distribution network dynamic topology reconstruction model is constructed by adopting a Bellman equation based on a Markov decision process tetrad, and instant cost and future cost are simultaneously considered in operation cost after topology reconstruction; and introducing an approximate dynamic programming method to transform the constructed dynamic topology reconstruction model of the power distribution network, and using a learning algorithm to complete model solution. According to the method, a stable power distribution network dynamic topology reconstruction scheme can be formulated under high-permeability photovoltaic power generation fluctuation, and the full-time operation cost is approximately optimal, so that the operation economy is improved, and new energy consumption is promoted.

Description

Technical Field

[0001] The present invention belongs to the technical field of power system operation optimization, and specifically relates to a method and system for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation. 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] Under large-scale distributed photovoltaic grid connection, the volatility of photovoltaic output reduces the economic efficiency of distribution network operation. It is necessary to use existing flexibility resources to resolve the contradiction between source-load uncertainty and the certainty of distribution network morphology and structure. As a special dispatchable resource, the topology of the distribution network can unleash the potential of the distribution network itself without additional investment by simply optimizing the network structure. With the large-scale application of intelligent distribution network automation switches, it is feasible to use the flexibility of the topology structure to achieve operation optimization. Topology reconstruction can change the topology by switching the line disconnection state according to the operating conditions, and manage the distribution of distribution network power flow through flexible topology changes, so as to achieve the goals of reducing losses, improving operation economy, and improving power supply reliability.

[0004] Traditional distribution networks have few distributed power sources and small load fluctuations. Topology reconstruction is often a long-term plan. Single static topology reconstruction is performed with load growth and seasonal load changes, and sequential reconstruction benefits are often not considered. However, distribution networks with high penetration of photovoltaic power generation have a large number of photovoltaic output fluctuations. The degree of fluctuation in a day is often greater than the worst fluctuation of a traditional distribution network over a considerable period of time. Therefore, it is necessary to make reconstruction operations in real time according to the fluctuation of photovoltaic power generation. Also, due to the large fluctuation of distributed photovoltaics, it is unrealistic to make topology reconstruction operations to adapt to a single time period. Using a greedy strategy to destroy the original topology structure to complete the reconstruction in each time period will lead to frequent changes in the line disconnection state, which will have an impact on the stable operation of the distribution network.

[0005] Although there are related adaptive reconstruction methods in the prior art, there are still problems:

[0006] First, the frequent fluctuations of photovoltaic power generation were not taken into account during the reconstruction process to ensure that the distribution network topology structure meets the optimal efficiency under photovoltaic fluctuations at all times. Second, user demand response, dispatchable photovoltaic response and energy storage equipment were not fully considered according to the dynamic topology reconstruction needs of the distribution network. Third, the established models are all optimization models under a single time section, which are essentially greedy strategies. They can only ensure the optimal efficiency of the reconstruction operation under a single time section, and fail to achieve sequential optimality at all times. Summary of the invention

[0007] In order to solve the above problems, the present invention proposes a method and system for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation, which are used to complete the acquisition of a dynamic topology reconstruction solution.

[0008] To achieve the above objectives, one or more embodiments of the present invention provide the following technical solutions:

[0009] In a first aspect, a method for dynamic topology reconstruction of a distribution network including high penetration photovoltaic power generation is disclosed, comprising the following steps:

[0010] Construct a Markov decision process quadruple;

[0011] A distribution network dynamic topology reconstruction model is established based on Markov decision process quaternary.

[0012] The constructed distribution network dynamic topology reconstruction model is transformed and solved to obtain the distribution network dynamic topology reconstruction solution.

[0013] As a further technical solution, the Markov decision process is modeled as a four-tuple (S, A, R, Pr), where S represents the state space, corresponding to the disconnected state of the dispatchable line; A represents the action space, corresponding to the reconstruction operation; R represents the reward function, corresponding to the operating cost after the reconstruction operation; Pr represents the state transition probability, corresponding to the selection probability of the reconstruction operation.

[0014] As a further technical solution, the objective function of the distribution network dynamic topology reconstruction model, i.e., the operating cost, includes immediate cost and future cost;

[0015] The immediate cost includes network reconstruction penalty, power purchase cost of the upper power grid, network loss cost, load demand response cost, photovoltaic demand response cost and energy storage battery charging and discharging cost; the future cost includes the cost in the subsequent period after the topology reconstruction plan is implemented at the current time point, which corresponds to the value function of the post-decision state introduced by the approximate dynamic programming method;

[0016] The constraints of the distribution network dynamic topology reconstruction model include operation constraints, safety limit constraints, radial topology constraints, energy storage battery charging and discharging state constraints, energy storage battery remaining capacity constraints, controllable photovoltaic constraints and controllable load constraints.

[0017] As a further technical solution, the distribution network dynamic topology reconstruction model is transformed using an approximate dynamic programming method:

[0018] The post-decision state is introduced to simplify the decision-making process and is used to distinguish the uncertainty before the decision and the uncertainty after the decision.

[0019] Assigning future costs to post-decision states transforms a multi-time-step large-scale model into a single-time-step deterministic model for each decision period;

[0020] Future closing cost and future disconnection cost are allocated to the opening and closing state of each dispatchable line, and an approximate value function based on linear basis function is designed to fit the value function of the state after decision making.

[0021] As a further technical solution, the solution process of the distribution network dynamic topology reconstruction model includes offline calculation and online calculation. The offline calculation is an iterative optimization process of multiple time steps, and the online calculation is a deterministic optimization process of a single time step.

[0022] As a further technical solution, offline calculation adopts time difference learning method for iterative solution. The iterative process uses the modified distribution network dynamic topology reconstruction model and adopts mixed integer programming method. The state value function after the decision at time t can be calculated using the value function observed at time t+1, and then the updated linear basis function weight value, i.e., the future closure cost and the future disconnection cost, is obtained. Through a sufficient number of iterations, the approximate value function of the future cost can be obtained.

[0023] As a further technical solution, online calculation uses the modified dynamic topology reconstruction model of the distribution network and adopts the mixed integer programming method. The approximate value function obtained by offline calculation is used to formulate a dynamic topology reconstruction plan for the distribution network according to the topological structure, network parameters, node load and photovoltaic prediction values ​​at each time point.

[0024] According to some embodiments, a second solution of the present invention provides a distribution network dynamic topology reconstruction system including high penetration photovoltaic power generation, which adopts the following technical solution:

[0025] In a second aspect, a dynamic topology reconstruction system for a distribution network including high penetration photovoltaic power generation is disclosed, comprising:

[0026] An acquisition module is configured to acquire a distribution network topology, network parameters, and node load and photovoltaic prediction values;

[0027] A modeling module, which is configured to construct a Markov decision process quadruple corresponding to the dynamic topology reconstruction problem and to construct a distribution network dynamic topology reconstruction model based on the quadruple;

[0028] The solution module is configured to solve the constructed distribution network dynamic topology reconstruction model based on the acquired information to obtain a distribution network dynamic topology reconstruction solution.

[0029] Among them, the objective function of the distribution network dynamic topology reconstruction model constructed by the modeling module, namely the operating cost, includes immediate cost and future cost; immediate cost includes network reconstruction penalty, power purchase cost of the upper power grid, network loss cost, load demand response cost, photovoltaic demand response cost and energy storage battery charging and discharging cost; future cost includes the cost in subsequent time periods after the topology reconstruction plan is implemented at the current time point, corresponding to the value function of the post-decision state introduced by the approximate dynamic programming method;

[0030] The solution module adopts the approximate dynamic programming method to transform the dynamic topology reconstruction model of the distribution network, introduces and allocates the future cost to the post-decision state, and designs an approximate value function based on the linear basis function to fit the value function of the post-decision state; the dynamic topology reconstruction model of the distribution network is solved by offline calculation and online calculation. The offline calculation adopts the time difference learning method and the mixed integer programming method to obtain the approximate value function. The online calculation uses the approximate value function obtained by offline calculation to obtain information and the mixed integer programming method to formulate the dynamic topology reconstruction plan of the distribution network.

[0031] According to some embodiments, a third solution of the present invention provides a computer-readable storage medium, which adopts the following technical solution:

[0032] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps in the method for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation as described in the first aspect of the present invention.

[0033] According to some embodiments, a fourth solution of the present invention provides an electronic device, which adopts the following technical solution:

[0034] An electronic device comprises a memory, a processor and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation as described in the first aspect of the present invention are implemented.

[0035] Compared with the prior art, the present invention has the following beneficial effects:

[0036] The present invention aims at the problem that the dynamic topology reconstruction of the distribution network needs to adapt to the volatility of high-penetration photovoltaic power generation. In order to ensure that each reconstruction operation meets the optimal operating cost under the photovoltaic fluctuations in the whole time period, a Markov decision process is adopted and the Bellman equation is used to establish a sequential dynamic topology reconstruction model that considers immediate cost and future cost. By considering the interaction between photovoltaic output changes, user demand response, dispatchable photovoltaic response and energy storage equipment, the immediate cost of the topology reconstruction operation in a single time step is calculated, and the future cost is calculated by considering the impact of each topology reconstruction operation on subsequent time periods. The sequential model is used to avoid the occurrence of short-sightedness effects, redundant operations, frequent reconstruction and other phenomena.

[0037] The present invention adopts an approximate dynamic programming method to introduce the post-decision state to transform the model, adopts a piecewise linear basis function to obtain an approximate value function of the future cost, uses time difference learning to update the future closing cost and the future disconnection cost in offline calculation, and uses the approximate value function to transform the sequential model into a single time step model in online calculation, completes the formulation of a dynamic topology reconstruction plan, and effectively solves the "curse of dimensionality" problem encountered in the process of solving the reconstruction plan caused by the large number of dispatchable lines in the distribution network, reduces the complexity of the model, and improves the solution speed. The dispatcher does not need to calculate the topological state that may be transferred at all time points, but only needs to update the approximate value function;

[0038] The present invention can complete the formulation of a dynamic topology reconstruction plan based on the approximate value function obtained by offline calculation, starting from any time point. Even if the topology structure does not change according to the predetermined plan due to events such as malfunction, temporary regulation and line failure, a new plan can be formulated using the topology structure at the next time point as the starting point using fast online calculation. BRIEF DESCRIPTION OF THE DRAWINGS

[0039] 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.

[0040] Figure 1 It is a flow chart of a method for dynamic topology reconstruction of a distribution network containing high penetration rate photovoltaic power generation in the first embodiment of the present invention;

[0041] Figure 2 is a schematic diagram of a Markov decision process in Embodiment 1 of the present invention;

[0042] Figure 3 Schematic diagram of the Markov decision process with the post-decision state introduced in the first embodiment of the present invention;

[0043] Figure 4 is a flow chart of solving a dynamic topology reconstruction model based on a time difference learning method in Embodiment 1 of the present invention;

[0044] Figure 5 Schematic diagram of the topological structure of the IEEE 33-node system in Embodiment 1 of the present invention;

[0045] Figure 6 It is a schematic diagram of the distribution network node load and photovoltaic prediction value in the first embodiment of the present invention;

[0046] Among them, (a) is a schematic diagram of load demand level, and (b) is a schematic diagram of photovoltaic output level;

[0047] Figure 7is a schematic diagram of the future closure cost iteration process in the first embodiment of the present invention; wherein,

[0048] (a) is a schematic diagram corresponding to the time step 7:00-8:00, (b) is a schematic diagram corresponding to the time step 8:00-9:00, (c) is a schematic diagram corresponding to the time step 9:00-10:00, (d) is a schematic diagram corresponding to the time step 10:00-11:00, (e) is a schematic diagram corresponding to the time step 11:00-12:00, and (f) is a schematic diagram corresponding to the time step 12:00-13:00;

[0049] Figure 8 is a schematic diagram of the future disconnection cost iteration process in the first embodiment of the present invention; wherein,

[0050] (a) is a schematic diagram corresponding to the time step 7:00-8:00, (b) is a schematic diagram corresponding to the time step 8:00-9:00, (c) is a schematic diagram corresponding to the time step 9:00-10:00, (d) is a schematic diagram corresponding to the time step 10:00-11:00, (e) is a schematic diagram corresponding to the time step 11:00-12:00, and (f) is a schematic diagram corresponding to the time step 12:00-13:00;

[0051] Fig. 9 This is a schematic diagram of the running results before and after the dynamic topology reconstruction of the small fluctuation scenario in the first embodiment of the present invention;

[0052] Fig.10 It is a structural diagram of a dynamic topology reconstruction system for a distribution network containing high-penetration photovoltaic power generation in the second embodiment of the present invention. DETAILED DESCRIPTION

[0053] The present invention will be further described below in conjunction with the accompanying drawings and embodiments.

[0054] It should be noted that the following detailed descriptions are all illustrative and 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.

[0055] It should be noted that the terms used herein are only for describing specific embodiments and are not intended to limit exemplary embodiments according to the present invention. As used herein, unless the context clearly indicates otherwise, the singular form is also intended to include the plural form. In addition, it should be understood that when the terms "comprising" and / or "including" are used in this specification, it indicates the presence of features, steps, operations, devices, components and / or combinations thereof.

[0056] In the absence of conflict, the embodiments of the present invention and the features of the embodiments may be combined with each other.

[0057] Based on the current status of the background technology, it is necessary to study a dynamic topology reconstruction strategy with high stability, considering whether each topology reconstruction action is applicable to the future period, that is, whether it has considerable benefits in the future period. Under this idea, the line disconnection command is issued cautiously, so that each reconstruction operation can ensure the economic operation under the fluctuation of photovoltaic output in the future period.

[0058] In view of the problem that dynamic topology reconstruction needs to adapt to the volatility of photovoltaic power generation, in order to ensure that each reconstruction operation meets the optimal benefit under the photovoltaic fluctuation of the whole period, a sequential dynamic topology reconstruction decision model is established by using the Markov decision framework and the Bellman equation; the interactive relationship between photovoltaic output changes, user demand response, dispatchable photovoltaic response and energy storage equipment in the distribution network is considered, and the immediate cost of the topology reconstruction operation under a single time step is calculated; the cost of the future period after each topology reconstruction operation is considered to avoid repeated operation of the same line and frequent reconstruction, and the future cost of the line opening and closing state in each reconstruction is approximately obtained by the approximate dynamic programming method; the post-decision state is introduced in the process to transform the Markov model, and the approximate value function of the future cost is obtained by using the piecewise linear basis function, and the future cost of the line state is updated in the offline training stage by using time difference learning. Finally, the sequential problem is transformed into a single time step decision problem by using the approximate value function to complete the formulation of the dynamic topology reconstruction strategy; the example simulation shows that the dynamic reconstruction method in this paper has more advantages in economy, is more flexible than the greedy strategy, and can avoid the short-sighted effect, and can ensure that the topology structure of the distribution network operation meets the optimal benefit of the whole period.

[0059] Embodiment 1

[0060] Embodiment 1 of the present invention introduces a method for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation.

[0061] like Figure 1 The flowchart of a method for dynamic topology reconstruction of a distribution network containing high penetration rate photovoltaic power generation is shown, comprising:

[0062] Step S01: obtaining the distribution network topology, network parameters, node load and photovoltaic prediction values ​​as the basic data of the power network for building a physical model;

[0063] Step S02: constructing a Markov decision process quadruple corresponding to the dynamic topology reconstruction problem;

[0064] Step S03: constructing a distribution network dynamic topology reconstruction model based on the Markov decision process quaternary;

[0065] Step S04: transforming and solving the constructed distribution network dynamic topology reconstruction model to obtain a distribution network dynamic topology reconstruction solution.

[0066] As one or more implementation modes, in step S01, topology structure identification, parameter state estimation, and node load and photovoltaic prediction are completed through distribution network automation equipment and node measurement devices.

[0067] As one or more implementation modes, in step S02, a Markov decision process quadruple is constructed corresponding to the dynamic topology reconstruction problem.

[0068] In the Markov decision process, actions are taken to transfer states according to the current state and expected rewards, and the ultimate goal is to maximize the long-term cumulative rewards. The core assumption of the Markov decision process is the Markov property, that is, the future state is only related to the current state and has nothing to do with the past state. This means that given the current state, the past state sequence has no effect on future decisions, which simplifies the modeling and solution of the problem. However, by considering the future benefits of the transferred state in the modeling, the model can be turned into sequential, thereby ensuring the optimal goal of the entire decision process. In the Markov decision process, the decision problem is modeled as a four-tuple (S, A, R, Pr). Among them, S represents the state space, including all possible states; A represents the action space, including all possible actions; R represents the reward function, which describes the immediate reward obtained under a given state and action; Pr represents the state transition probability, which describes the probability distribution of transitioning to the next state under a given state and action. Figure 2 The Markov decision process is shown. The decision maker observes the current state S1 at time t1, and performs action A1 according to a specific strategy based on the current state and S2 environmental observation, and obtains a reward R1. At t2, based on the current state S2 and S3 environmental observation, action A2 is performed according to a specific strategy, and a reward R2 is obtained. At the last moment t3, based on the current state S3 and the terminal state, action A3 is performed according to a specific strategy, and a reward R3 is obtained. If future costs are considered in the reward, the future costs at the last moment usually become fixed values.

[0069] Step S201: Determine the state space S.

[0070] Determine the state space S, which is used to construct the Markov decision process quadruple. This is the basic step of the Markov method. Only after the state is determined can the value function of the state be obtained.

[0071] Corresponding to the topology reconstruction problem, the distribution network state is composed of the topological structure of each time step, and the state space corresponding to each moment is:

[0072] S t ={E t (X t,12 ,X t,23 ,...,X t,in ,...)}in∈L (1)

[0073] Among them, E t is the topological structure of the distribution network at time t; n is the node number, the total number of nodes is N; X t,in is the open / closed state of line in at time t, 0 means open and 1 means closed; L is the set of all lines, corresponding to the total number of lines.

[0074] Step S202: Determine the action space A.

[0075] Determine the action space A, which is used to construct the Markov decision process quadruple. This is the basic step of the Markov method, which is used to reflect the transition process between two states.

[0076] Action Space A t Indicates that from the current state S t The topology reconstruction actions that may be taken at the next moment.

[0077] The decision maker determines the optimal action a based on the current state and the predicted value of the future environment. t , ensuring that the expected value of the reward is optimal. In this process, the action is a function relative to the current observed state and the possible transition state:

[0078] a t =π(S t ,S t+1 ) t ∈A t (2)

[0079] Among them, S t is the current state; S t+1 is a set of possible transition states; π(·) is a specific deterministic action selection strategy. Since the distribution network topology of the transition state is directly determined by the action, E t+1 By action a t The only decision is that a deterministic state transition will occur after the action is executed.

[0080] Step S203: Determine the reward function R.

[0081] In the formulation of the dynamic topology reconstruction plan of the distribution network, the operation and control capabilities of flexible loads and dispatchable photovoltaics are utilized, and the flexible operation potential of the distribution network topology is explored through the dynamic topology reconstruction strategy, the operation cost is reduced, and the social benefit is optimized. The specific reward function is given in step S03, and an objective function including network reconstruction penalty, power purchase cost of the upper power grid, network loss cost, load demand response cost, photovoltaic demand response cost, energy storage battery charging and discharging cost and future cost is formulated to form a single-objective optimization problem and reduce the complexity of the model.

[0082] The reward function is the objective function of formula (4) in the model.

[0083] Step S204: Determine the state transition probability Pr.

[0084] The state transition probability describes the probability of transitioning to the next state given the current state and taking an action in the Markov decision process. The state transition probability function is:

[0085]

[0086] Among them, Pr(S t+1,h |S t,h' ,a t ) is based on action a t , from a specific Markov state S at time t t,h' Transfer to S t+1,h The first term on the right is the probability of action a t , the topology is from E t Transfer to E t+1 probability.

[0087] According to the analysis in step S202, a deterministic Markov decision process is established.

[0088] The state transition probability is used to illustrate that a deterministic state transition has been constructed. The subsequent consideration of whether the reconstruction action has the probability of transferring to a different state is no longer considered. That is, the state after the line switch operation is determined. This is a conventional step in the Markov method.

[0089] As one or more implementation methods, in step S03, a distribution network dynamic topology reconstruction model is established based on the Markov decision process framework. A sequential model is established, which takes into account the cost of the entire period, and is more conducive to judging the impact of the reconstruction operation on the cost of the current period and future periods, and formulating a better reconstruction plan.

[0090] When formulating topology reconstruction actions at different times, the decision results not only affect the current operating costs, but also the future operating costs. Therefore, when formulating the dynamic topology reconstruction model, the objective function is established using the Bellman equation, and the state value function is used and decomposed into two parts: immediate cost and future cost to form a sequential model. The formulation of the operating cost aims to achieve the optimal social benefit and ensure the optimal cost of the dynamic topology reconstruction solution at all times.

[0091] The objective function is:

[0092]

[0093] Among them, v t (S t,h' ) and v t+1 (S t+1,h ) are the value functions at the corresponding moments; C t(a t ) is to take action a t The immediate reward after t+1 includes the network reconstruction penalty and the distribution network operation cost in the t+1 period, as shown in formula (5):

[0094] shown.

[0095] C t (a t )=C t,sw +C t,sub +C t,loss +C t,lr +C t,pv +C t,bat (5)

[0096] Among them, C t,sw is the network reconstruction penalty; C t,sub is the electricity purchase cost of the upper power grid; C t,loss is the network loss cost; C t,lr is the load demand response cost; C t,pv is the PV demand response cost; C t,bat It is the cost of charging and discharging the energy storage battery.

[0097] (1) Network Reconfiguration Penalty

[0098]

[0099] Among them, α(n) represents the previous node connected to node n along the radial topology, and the root node n = 1; f in It is a line disconnection penalty.

[0100] (2) Power purchase cost of the upper power grid

[0101] C t,sub =P t+1,sub c sub (7)

[0102] Among them, P t+1,sub is the amount of electricity purchased by the upper power grid, specifically the active power delivered by the root node; c sub It is the unit electricity price.

[0103] (3) Network loss cost

[0104]

[0105] Among them, I t+1,in is the line current; r in is the line resistance from i to n; c loss is the unit network loss cost.

[0106] (4) Load demand response cost

[0107]

[0108] Among them, P t+1,n,load is the load forecast value; η n is the dispatchable load ratio of n nodes; μ t+1,n,lr Yes

[0109] The load demand response ratio of the node; c lr is the demand response cost per unit load.

[0110] (5) Photovoltaic demand response costs

[0111]

[0112] Among them, P t+1,n,pv is the predicted value of dispatchable PV capacity; μ t+1,n,pv is the PV demand response ratio of n nodes; c pv is the unit PV demand response cost.

[0113] (6) Charging and discharging costs of energy storage batteries

[0114]

[0115] Among them, E t+1,n is the remaining capacity of the energy storage after the end of time step t+1; c bat It is the unit charging and discharging cost of energy storage.

[0116] The action space between the two states is very large. Assuming that the distribution network has n dispatchable lines and does not consider the radial topology constraint, the distribution network has 2 n Topological structure, that is, the action space contains 2 n There are actions, each of which has a selection probability. According to step S202, the action selection strategy π collapses the selection probability distribution of the action into a deterministic action, so that the topology reconstruction strategy of each time step can be selected in turn:

[0117]

[0118] In the decision-making process, the operation and reconstruction of the distribution network need to meet actual constraints. The constraints of the dynamic topology reconstruction model of the distribution network are as follows:

[0119] (1) Operation constraints

[0120] The Distflow model, an accurate AC power flow equation suitable for radial distribution networks, is used to model the operation constraints. The optimization variables selected for the constraints include node voltage, line current, node active and reactive power, and line active and reactive power. The power flow relationship is:

[0121]

[0122] Where β(n) represents the subsequent nodes connected to node n along the radial topology; I 2,t+1,in is the square of the line current; P t+1,nj and P t+1,in are the power sent from node n to j and the power sent from node i to n respectively; x in is the line reactance from node i to n; P t+1,n is the active input power of node n itself, as shown in formula (15); Q t+1,n is the reactive input power of node n itself, as shown in formula (16).

[0123]

[0124] in, and are the charging and discharging powers of the energy storage at the t+1 time step respectively.

[0125] The node voltage drop relationship is as follows. The Big-M method is used to ensure that the voltage drop relationship does not force the voltage amplitudes at both ends of the unconnected branches to be equal when the branch is disconnected:

[0126] m in =(1-X t+1,in )M (17)

[0127] Among them, M is a large positive number.

[0128]

[0129] The relationship between branch current and voltage, line active power, and line reactive power is relaxed to a second-order cone form:

[0130]

[0131] Among them, V 2,t+1,n is the square of the voltage at node n, and the voltage amplitude at the root node n=1 is 1, as shown in formula (20).

[0132]

[0133] (2) Safety restrictions and constraints

[0134] The node voltage and line current of the distribution network are subject to operating limitations. In actual operation and regulation, it is necessary to ensure that the voltage and current do not exceed the limits. This is even more important in distribution networks with high penetration rates of photovoltaic power generation.

[0135]

[0136]

[0137] Among them, V n,min and V n,max are the upper and lower limits of the voltage value of node n respectively; I in,max is the upper limit of the line in current value. In addition, the root node peak load depth limit needs to be met:

[0138] κP n,sub,max <P t+1,n,sub <P n,sub,max (twenty three)

[0139] Among them, P n,sub,max is the maximum power of the root node; κ is the maximum peak-shaving depth proportional coefficient of the root node.

[0140] (3) Radial topology constraints

[0141] When the distribution network is operating normally, the topology structure should be radial. When formulating the topology reconstruction strategy, the radial constraint requirements of the network must be met to ensure that there are no isolated nodes in the network after reconstruction. The radial topology constraints are:

[0142]

[0143] For the operation without isolated nodes, equation (25) is sufficient to ensure the radial shape of the distribution network. However, the distribution network contains high penetration of photovoltaic power generation. When the distributed generation and load value in a certain area are the same, an island will appear in the model. Therefore, an auxiliary flow is introduced to ensure the connectivity of the network. The auxiliary flow does not affect the flow distribution in the model, but only serves as a constraint to ensure that the line disconnection state meets the radial topology constraint:

[0144]

[0145] in, and It is the auxiliary current on the line; is the maximum value of the auxiliary power flow on the line;

[0146] is a smaller node auxiliary injection power value.

[0147] (4) Constraints on the charging and discharging status of energy storage batteries

[0148] The energy storage battery maintains the same charge and discharge state in the same time step, and the charge and discharge power is affected by the battery properties.

[0149] Can limit:

[0150]

[0151] in, and They are the charging and discharging states of the energy storage at node n; and are the maximum charging and discharging power of energy storage respectively.

[0152] (5) Constraints on remaining capacity of energy storage batteries

[0153]

[0154] 0≤E t+1,n ≤E t+1,n,max (31)

[0155] Among them, u ch and u dis They are the charging and discharging efficiency of energy storage, which is a constant between 0 and 1; E t+1,n,max

[0156] It is the maximum capacity limit of energy storage.

[0157] (6) Controllable photovoltaic constraints

[0158]

[0159] 0≤μ t+1,n,pv ≤1 (33)

[0160] in, is the power factor angle limit of dispatchable photovoltaic.

[0161] (7) Controllable load constraints

[0162] 0≤μ t+1,n,lr ≤1 (34)

[0163] As one or more implementation modes, in step S04, the constructed distribution network dynamic topology reconstruction model is modified and solved to obtain a distribution network dynamic topology reconstruction solution.

[0164] Step S401: introducing a post-decision state transformation model.

[0165] Approximate dynamic programming methods combine dynamic programming principles with machine learning techniques to approximate solutions to decision problems whose state space is too large and complex for traditional dynamic programming. For example, the state space corresponding to the reconstruction action in the dynamic topology reconstruction problem is too large. Among them, the post-decision state is one of the means by which the approximate dynamic programming method processes the model, which helps to simplify and optimize the decision-making process. It refers to the state of the system after the action is taken and before the next decision time arrives, reflecting the direct result of the action execution. By separating the impact of the decision from the random events that may occur subsequently, the problem modeling is made more intuitive, thereby simplifying the structure of the state space and the subsequent decision-making process.

[0166] like Figure 3As shown, S 1,A , S 2,A , S 3,A The post-decision states after S1, S2, and S3 take actions A1, A2, and A3 are respectively, and the future cost is allocated to the post-decision state, thereby transforming the multi-time-step large-scale Markov decision process model into a single-time-step deterministic model. t Status S t,h' For example, the post-decision state uses It means that formula (4) can be expressed as the expected form:

[0167]

[0168] Then, using the post-decision state transformation formula (35), we obtain a single time step deterministic model:

[0169]

[0170] in, is the value of the state after the decision, and then:

[0171]

[0172] Substituting formula (36) into formula (37), we get:

[0173]

[0174] Therefore, the post-decision state value is expressed as formula (39). The post-decision state value can be calculated based on the state quantity of the decision process at the next moment, and then the approximate value function method is used to fit the post-decision state value.

[0175]

[0176] Step S402: construct an approximate value function to fit the future cost.

[0177] The main difference between approximate dynamic programming and dynamic programming is that the training process is introduced to obtain the approximate value function, and the approximate value function is used instead of the exact value to deal with the "curse of dimensionality" problem caused by the excessive state space and action space, and to solve the optimal solution of the multi-period problem. For the approximate value function, the commonly used methods can be summarized into three categories, namely the lookup table method, the parameter model method and the parameter-free model method. Among them, the lookup table method is not suitable for the dynamic topology reconstruction problem of the distribution network with a large number of topological structure states. It is very difficult to maintain the lookup table for a large number of feasible topological structures; the parameter-free model requires a large amount of sample data for training, and the model has poor interpretability. It is also not suitable for the dynamic topology reconstruction plan formulation work carried out recently; in comparison, the biggest advantage of the parameter model method is that it uses a model with a fixed number of parameters to obtain the approximate value function, which can be effectively approximated even in the face of a large state space. The approximate value function is constructed using a basis function model that is easy to train and has strong generalization ability. Specifically, a linear basis function model is selected, and the form is as follows:

[0178]

[0179] in, is the approximation of the state value function after the decision; φ r is the selected linear basis function; θ t,r is the weight value of the corresponding basis function; R is the basis function set.

[0180] The decision variable of the dynamic topology reconstruction model of the distribution network is the disconnection state of the dispatchable line. Therefore, the line disconnection state cost is used to set the linear basis function. The specific linear basis function model is:

[0181]

[0182] Among them, c t,l,on and c t,l,off They are the future closure cost and future disconnection cost of the line respectively.

[0183] Step S403: Offline calculation using a temporal difference learning method to update the approximate value function.

[0184] The temporal difference learning method is a reinforcement learning method that combines the Monte Carlo method with the idea of ​​dynamic programming. It does not require prior knowledge of the model, that is, it does not need to know the reward function, but learns directly from experience. The core is to use the observed state value to update the approximate value function of the current state. According to formula (39), the state value function after the decision at time t can be calculated using the value function observed at time t+1. The temporal difference learning method is used to iteratively update the value function of the state after the decision. The offline calculation is based on Figure 4 The update formula is expressed as:

[0185]

[0186] in, is the approximate state value of the post-decision state at the yth iteration; is the state value observed at the yth iteration t+1; ε is the learning coefficient. Through enough iterations, the converged post-decision state approximation function can be obtained for solving the single time step deterministic model.

[0187] According to formula (41), the approximate value function update strategy corresponding to formula (42) is transformed into:

[0188]

[0189] in, It is the update coefficient of future cost. Lines with higher future operating costs correspond to relatively higher future closure costs, that is, corresponding to higher cost value updates.

[0190] In order to add more exploration mechanisms in the iteration process, the ε-greedy strategy is used to balance the exploration and utilization process. In the early stage of iteration, exploration is increased to gain a more comprehensive understanding of the state space; as the number of iterations increases, exploration is gradually reduced to make more use of the known best solution. Therefore, the random coefficient ε is set and the c ε Decays with the number of iterations.

[0191] ε y =ε y-1 c ε (44)

[0192] where ε is the probability of randomly assigning a topology at each iteration.

[0193] Step S404: Online calculation and use of approximate value functions to formulate a dynamic topology reconstruction plan.

[0194] The future cost is calculated by the approximate value function obtained in step S403, and the single time step distribution network dynamic topology

[0195] The reconstructed model is:

[0196]

[0197] Formula (12) is transformed into:

[0198]

[0199] The solution process is the temporal difference learning method, which is a commonly used method and is described in detail in step S403. The solution process description in this paragraph is the same as the steps formulated in S403.

[0200] The commercial solver cplex is used to solve the model according to formula (45). The output is the line state S at each moment in the objective function of formula (4), which is the reconstruction plan.

[0201] Case Analysis

[0202] The equipment configuration in this example is as follows Figure 5 As shown in the figure, the photovoltaic equivalent capacity of distributed photovoltaic nodes is 100, 100, 100, 300, 300, 100, 100 and 100kW according to the node number, and the dispatchable photovoltaic installed capacity is 200, 400, 400 and 200kW according to the node number. The greed coefficient is initialized to 0.35 and decays according to 0.978.

[0203] In order to explore the adaptability of the dynamic topology reconstruction method of the distribution network to the fluctuation of high-penetration photovoltaic output, two scenarios are set for result calculation, namely the small fluctuation scenario of photovoltaic power generation and the large fluctuation scenario. In the small fluctuation scenario, the cumulative daily power generation of the configured photovoltaic power generation is as high as 47.48%, which is in line with the setting of the high-penetration photovoltaic distribution network scenario. The load and photovoltaic output level prediction values ​​under each scenario are as follows: Figure 6 As shown, (a) schematic diagram of load demand level, (b) schematic diagram of photovoltaic output level.

[0204] The calculation example takes 7:00 as the initial decision state, and the topology structure is that all segment switches are closed and all tie switches are open. The transition state time range is 7:00-19:00, and the future cost after the last transition state is set to the cost of returning to the initial decision state at 19:00. For the immediate reward part, the topology reconstruction penalty coefficient is set to 1, c t+1,sub Set to 0.55, c loss Set to 2.0, c lr Set to 1.0, c pv Set to 1.5, c bat Set to 0.7, the maximum peak load regulation depth of the root node is 30%, the rated power is 4000kW, and the node load response ratio η n Set to 0.2, the maximum energy storage of the energy storage battery is 400 and 800kW respectively, and the capacity in the initial state is 1 / 4 of the maximum energy storage.

[0205] (1) Scenarios where photovoltaic output fluctuates slightly

[0206] The maximum number of iterations of offline calculation is set to 240, the future closing cost and future disconnection cost of the initial state are initialized to 0.03, the learning coefficient is set to 0.08, the cost update coefficient is set to 0.4, and the approximate value function of the future cost is updated through the offline calculation process. The average offline calculation time in all scenarios is maintained at 4.15h, and the online calculation time is 59s. The cost convergence of the first 6 time steps is shown as an example. Figure 7 and Figure 8 As shown in the figure, it can be found that both the future closing cost and the future disconnection cost have inflection points that violate the original update direction, which correspond to the new optimal dynamic topology reconstruction path mined by the time difference learning method through long-term learning. The proposed method can consider the future cost of topology reconstruction actions at different time steps and continuously improve the accuracy of the approximate value function through time difference learning.

[0207] Figure 7 Schematic diagram of the future closure cost iteration process, where (a) is the schematic diagram corresponding to the time step 7:00-8:00, (b) is the schematic diagram corresponding to the time step 8:00-9:00, (c) is the schematic diagram corresponding to the time step 9:00-10:00, (d) is the schematic diagram corresponding to the time step 10:00-11:00, (e) is the schematic diagram corresponding to the time step 11:00-12:00, and (f) is the schematic diagram corresponding to the time step 12:00-13:00.

[0208] Figure 8 Schematic diagram of the future disconnection cost iteration process in the first embodiment of the present invention, wherein (a) is a schematic diagram corresponding to the time step 7:00-8:00, (b) is a schematic diagram corresponding to the time step 8:00-9:00, (c) is a schematic diagram corresponding to the time step 9:00-10:00, (d) is a schematic diagram corresponding to the time step 10:00-11:00, (e) is a schematic diagram corresponding to the time step 11:00-12:00, and (f) is a schematic diagram corresponding to the time step 12:00-13:00.

[0209] The comparison results of the operating costs of using the proposed method and maintaining the initial topology structure in a small fluctuation scenario are as follows: Fig. 9 As shown in the figure. After adopting the proposed method, the operating cost is reduced by 462.54 yuan in total, and the social benefit is effectively improved by 3.30%. The operating cost is reduced to a certain extent in each time step. Given that the total load value of the example taken is small, applying this benefit improvement ratio to the actual large number of distribution networks with available topology reconstruction resources will save considerable operating costs. The proposed method is compared with the greedy strategy that only performs instant cost optimization, and the line changes are shown in Table 1.

[0210] Table 1 Comparison of reconstruction results for small fluctuation scenarios

[0211]

[0212] From Table 1, it can be found that the proposed method can obtain a better operating cost than the greedy strategy by completing only one topology reconstruction in the initial state. Sequential topology reconstruction can avoid repeated disconnection of the line in the period of 14:00-17:00 in actual operation. Smaller PV output fluctuations will not affect the stability of the line state, because in the long run, high-frequency topology reconstruction is only to obtain the optimal single time step in a small range of optimization intervals using the topology reconstruction penalty, but it may not be advantageous from the perspective of long-term benefits.

[0213] (2) Scenarios where photovoltaic output fluctuates significantly

[0214] A scenario of large PV fluctuations is set to verify whether the proposed method will be affected by the frequent fluctuations of distributed PV and formulate redundant topology reconstruction schemes. The comparison of the running results of the proposed method and the greedy strategy in this scenario is shown in Table 2.

[0215] Table 2 Comparison of reconstruction results for large fluctuation scenarios

[0216]

[0217]

[0218] From Table 2, it can be found that the proposed method can better adapt to the volatility of distributed photovoltaic output and formulate a stable topology reconstruction strategy. When the future cost-benefit of photovoltaic output changes over a long period of time cannot offset the reconstruction penalty, no redundant reconstruction actions will be made. Compared with the greedy strategy, the number of line disconnection state changes is reduced by nearly half, and the reconstruction actions are also reduced by 5 times. At the same time, the greedy strategy becomes more unstable in the scenario of large fluctuations, and more serious repeated reconstruction actions occur in the period of 7:00-11:00. By adjusting the reconstruction penalty coefficient, Table 3 is obtained for sensitivity verification.

[0219] Table 3 Scenarios with different reconstruction penalty coefficients

[0220]

[0221] The topology reconstruction penalty coefficient corresponds to the frequency of dynamic topology reconstruction strategies required by dispatchers. The higher the penalty coefficient, the more stable the reconstruction strategy, which means that dispatchers are less willing to perform topology reconstruction operations. Table 3 shows that the short-sighted effect of the greedy strategy becomes more obvious as the topology reconstruction penalty increases. The proposed method can find a better future cost from the penalty value. When the penalty coefficient is 1 and 2, dispatchers are allowed to perform more frequent topology reconstruction operations. Compared with the greedy strategy, the proposed method has obvious stability advantages and can avoid most invalid reconstructions. When the penalty coefficient is 3, 4, and 5, the controller needs to reduce the reconstruction operation, but still wants to use topology reconstruction to achieve a certain degree of economic operation. The greedy strategy can only observe the immediate cost part. When the immediate cost cannot offset the reconstruction penalty, it will fall into the local optimum and cannot find the optimal topology path. However, the proposed method can observe the impact of the current operation on the subsequent time period. The reconstruction operation of some time periods may not meet the optimal immediate cost, but it can ensure that the future cost will decrease to a certain extent.

[0222] This example verifies the adaptability of the proposed method to fluctuations in high-penetration photovoltaic power generation by setting up two scenarios. The proposed method has complete advantages over the greedy strategy in the scenario of small photovoltaic fluctuations. The proposed method in the scenario of large fluctuations can still achieve better social benefit optimization effects than the greedy strategy, which proves the effectiveness of the dynamic reconstruction method of the distribution network, ensures that the reconstruction plan meets the optimal operating cost throughout the entire period, and achieves the purpose of reducing the operating cost of the distribution network, adapting to photovoltaic power generation fluctuations and promoting the consumption of new energy.

[0223] Embodiment 2

[0224] Embodiment 2 of the present invention introduces a dynamic topology reconstruction system for a distribution network including high-penetration photovoltaic power generation.

[0225] like Fig.10 A distribution network dynamic topology reconstruction system with high penetration rate photovoltaic power generation is shown, comprising:

[0226] An acquisition module is configured to acquire a distribution network topology, network parameters, and node load and photovoltaic prediction values;

[0227] A modeling module, which is configured to construct a Markov decision process quadruple corresponding to the dynamic topology reconstruction problem and to construct a distribution network dynamic topology reconstruction model based on the quadruple;

[0228] A solution module, which is configured to solve the constructed distribution network dynamic topology reconstruction model based on the acquired information to obtain a distribution network dynamic topology reconstruction solution;

[0229] Among them, the objective function of the distribution network dynamic topology reconstruction model constructed by the modeling module, namely the operating cost, includes immediate cost and future cost; immediate cost includes network reconstruction penalty, power purchase cost of the upper power grid, network loss cost, load demand response cost, photovoltaic demand response cost and energy storage battery charging and discharging cost; future cost includes the cost in subsequent time periods after the topology reconstruction plan is implemented at the current time point, corresponding to the value function of the post-decision state introduced by the approximate dynamic programming method;

[0230] The solution module adopts the approximate dynamic programming method to transform the dynamic topology reconstruction model of the distribution network, introduces and allocates the future cost to the post-decision state, and designs an approximate value function based on the linear basis function to fit the value function of the post-decision state; the dynamic topology reconstruction model of the distribution network is solved by offline calculation and online calculation. The offline calculation adopts the time difference learning method and the mixed integer programming method to obtain the approximate value function. The online calculation uses the approximate value function obtained by offline calculation to obtain information and the mixed integer programming method to formulate the dynamic topology reconstruction plan of the distribution network.

[0231] The detailed steps are the same as those of the method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation provided in Example 1, and will not be repeated here.

[0232] Embodiment 3

[0233] Embodiment 3 of the present invention provides a computer-readable storage medium.

[0234] A computer-readable storage medium stores a program thereon, which, when executed by a processor, implements the steps in the method for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation as described in Embodiment 1 of the present invention.

[0235] The detailed steps are the same as those of the method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation provided in Example 1, and will not be repeated here.

[0236] Embodiment 4

[0237] A fourth embodiment of the present invention provides an electronic device.

[0238] An electronic device comprises a memory, a processor and a program stored in the memory and executable on the processor, wherein when the processor executes the program, the steps in the method for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation as described in Embodiment 1 of the present invention are implemented.

[0239] The detailed steps are the same as those of the method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation provided in Example 1, and will not be repeated here.

[0240] Embodiment 5

[0241] 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.

[0242] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. For those skilled in the art, the present invention may have various modifications and variations. Any modification, equivalent replacement, improvement, etc. made within the spirit and principle of the present invention shall be included in the protection scope of the present invention.

[0243] 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 dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation, characterized in that: include: Construct a Markov decision process quadruple; A distribution network dynamic topology reconstruction model is established based on Markov decision process quaternary. The constructed distribution network dynamic topology reconstruction model is transformed and solved to obtain the distribution network dynamic topology reconstruction solution.

2. A method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation according to claim 1, characterized in that: The Markov decision process is modeled as a four-tuple (S, A, R, Pr), where S represents the state space, corresponding to the disconnection state of the dispatchable line; A represents the action space, corresponding to the reconstruction operation; R represents the reward function, corresponding to the operating cost after the reconstruction operation; Pr represents the state transition probability, which corresponds to the selection probability of the reconstruction operation.

3. A method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation according to claim 1, characterized in that: The objective function of the distribution network dynamic topology reconstruction model, i.e., the operating cost, includes immediate cost and future cost; The immediate cost includes network reconstruction penalty, power purchase cost of the upper power grid, network loss cost, load demand response cost, photovoltaic demand response cost and energy storage battery charging and discharging cost; the future cost includes the cost in the subsequent period after the topology reconstruction plan is implemented at the current time point, which corresponds to the value function of the post-decision state introduced by the approximate dynamic programming method; The constraints of the distribution network dynamic topology reconstruction model include operation constraints, safety limit constraints, radial topology constraints, energy storage battery charging and discharging state constraints, energy storage battery remaining capacity constraints, controllable photovoltaic constraints and controllable load constraints.

4. A method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation according to claim 1, characterized in that: The distribution network dynamic topology reconstruction model is transformed using an approximate dynamic programming method: The post-decision state is introduced to simplify the decision-making process and is used to distinguish the uncertainty before the decision and the uncertainty after the decision. Assigning future costs to post-decision states transforms a multi-time-step large-scale model into a single-time-step deterministic model for each decision period; Future closing cost and future disconnection cost are allocated to the opening and closing state of each dispatchable line, and an approximate value function based on linear basis function is designed to fit the value function of the state after decision making.

5. A method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation according to claim 1, characterized in that: The solution process of the distribution network dynamic topology reconstruction model includes offline calculation and online calculation. The offline calculation is an iterative optimization process of multiple time steps, and the online calculation is a deterministic optimization process of a single time step.

6. A method for dynamic topology reconstruction of a distribution network containing high penetration photovoltaic power generation according to claim 1, characterized in that: The offline calculation adopts the time difference learning method for iterative solution. The iterative process uses the modified distribution network dynamic topology reconstruction model and adopts the mixed integer programming method. The state value function after the decision at time t can be calculated using the value function observed at time t+1, and then the updated linear basis function weight value, i.e., the future closing cost and the future disconnection cost, can be obtained. With a sufficient number of iterations, the approximate value function of the future cost can be obtained. The online calculation uses the modified dynamic topology reconstruction model of the distribution network and adopts the mixed integer programming method. The approximate value function obtained by offline calculation is used to formulate a dynamic topology reconstruction plan for the distribution network according to the topological structure, network parameters, node load and photovoltaic prediction values ​​at each time point.

7. A dynamic topology reconstruction system for a distribution network with high penetration photovoltaic power generation, characterized in that: include: An acquisition module is configured to acquire a distribution network topology, network parameters, and node load and photovoltaic prediction values; A modeling module, which is configured to construct a Markov decision process quadruple corresponding to the dynamic topology reconstruction problem and to construct a distribution network dynamic topology reconstruction model based on the quadruple; The solution module is configured to solve the constructed distribution network dynamic topology reconstruction model based on the acquired information to obtain a distribution network dynamic topology reconstruction solution.

8. A computer-readable storage medium having a program stored thereon, characterized in that: When the program is executed by a processor, the steps in the method for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation as described in any one of claims 1 to 6 are implemented.

9. An electronic device comprising a memory, a processor, and a program stored in the memory and executable on the processor, characterized in that: When the processor executes the program, the steps in the method for dynamic topology reconstruction of a distribution network containing high-penetration photovoltaic power generation are implemented as described in any one of claims 1-6.

10. A computer program product, comprising a computer program, characterized in that When the computer program is executed by a processor, the method according to any one of claims 1 to 6 is implemented.

Images