Intelligent soft switching method for highway active power distribution network operation scheduling

Through the multivariate uncertain data model and intelligent soft switch technology based on Wasserstein distance, the 'traffic-electrical' coupling network of the highway ADN system was built, which solved the problem of power grid interconnection and power exchange in the dynamic time domain of the ADN system, and realized the system The improvement of safety and economics has slowed down the fluctuations of the power grid.

CN118920601APending Publication Date: 2024-11-08LANZHOU JIAOTONG UNIV +1
View PDF 4 Cites 0 Cited by

Patent Information

Application Number
CN202410954290.7
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2024-07-17
Publication Date
2024-11-08

AI Technical Summary

Technical Problem

The existing highway ADN system is difficult to realize dynamic interconnection and power exchange of the power grid in the dynamic time domain, and is affected by the uneven installation location and installed capacity of power generation equipment, resulting in a large number of lines and switches, a large variety of access equipment, and a complex model, making it difficult to Give full play to initiative.

Method used

Using a multivariate uncertain data model based on Wasserstein distance, a robust set and dynamic Wasserstein uncertain data set are constructed, combined with intelligent soft switch (SOP) equipment, a 'traffic-electrical' coupled ADN network is constructed, and optimized through NSGA-II and PSO algorithms SOP switching state and power scheduling realizes the second-order optimization model of the ADN system.

Benefits of technology

It improves the information interaction and power transmission capabilities between the power grid of the ADN system, improves the voltage level, enhances the safety and economy of the system, and realizes effective suppression of the power grid fluctuations caused by EV charging load and grid connection.

✦ Generated by Eureka AI based on patent content.

Smart Images

  • Figure CN118920601A_ABST
    Figure CN118920601A_ABST
Patent Text Reader

Abstract

The invention discloses an intelligent soft switching method for operation scheduling of an expressway active power distribution network, relates to the technical field of electric energy scheduling of electrified expressways, and constructs an ADN model with the optimization objectives of minimizing the comprehensive power consumption cost and maximizing the robustness of an ADN system. The construction of an expressway ADN system and the robust optimization and cost optimization work of a power system under multi-objective optimization are solved and realized by combining a specific second-order multi-objective optimization model and introducing a double-layer heuristic algorithm, the power supply elasticity of an expressway power supply system facing extreme scenes is improved, and the flexible allocation degree of electric energy is further realized. The optimization model and the solving method are based on an advanced mathematical optimization algorithm, flexible dispatching of expressway electric energy can be achieved in the environment of multi-element uncertain parameter fluctuation, the degree of participation of a distributed power grid is improved to the maximum extent, dependence on traditional energy is reduced, and a power system is promoted to be transformed into clean energy.
Need to check novelty before this filing date? Find Prior Art

Description

Technical Field

[0001] The invention relates to the technical field of electric energy dispatching of electrified highways, and in particular to an intelligent soft switching method for operation dispatching of active power distribution networks of highways. Background Art

[0002] At present, scholars at home and abroad are committed to studying ADN that can effectively integrate dispatchable resources in the network, trying to build a highway ADN system with high flexibility and high reliability by connecting the power grids along the highway in series to achieve the purpose of weakening power flow fluctuations. However, due to the uneven installation location and installed capacity of power generation equipment, linking ADN devices to form a new form of distribution network will lead to a large number of lines and switches, a large number of access devices, and a complex model construction. How to give full play to the initiative of ADN in the dynamic time domain and realize dynamic interconnection and power exchange of power grids is the main problem facing ADN. The information and power interaction between independent power grids is mainly achieved through distributed power sources, intelligent soft switches (Soft open point, SOP) and energy storage devices. Formulating appropriate SOP installation locations and disconnection plans can flexibly change the power grid structure and improve its electrical characteristics.

[0003] Based on the above problems, the present invention constructs an intelligent soft switch technology for the operation and dispatching of the active distribution network of highways. By using SOP equipment to connect the long-chain ADN system of the independent power grid along the highway in series, the power supply of the entire highway section can be realized. In order to further ensure the stability and economy of the power supply of highways, a second-order optimization model adapted to the power dispatching of the ADN system is proposed. The electricity cost is minimized under the premise of trying to ensure the overall stability and optimization of the ADN. The model can effectively smooth the grid fluctuations caused by EV charging load and DM grid connection. Summary of the invention

[0004] The purpose of the present invention is to solve the above-mentioned problems and to provide an intelligent soft switching method for operation and scheduling of active distribution networks on highways.

[0005] To achieve the above object, the technical solution adopted by the present invention is:

[0006] According to one aspect of the present invention, there is provided an intelligent soft switching method for operation scheduling of active distribution network of highway, which is used for power scheduling of electrified highway, and is characterized in that it comprises the following steps:

[0007] S1. Highway multivariate uncertain data model based on Wasserstein distance;

[0008] S1.1, Wasserstein distance definition and its application characteristics;

[0009] The definition of Wasserstein distance involves all possible joint distributions between two probability distributions, and the final distance value is obtained by taking the lower bound of the expected value of the distance between the samples under these joint distributions. Specifically, for two probability distributions P and P K , its Wasserstein distance is defined as:

[0010]

[0011] In the formula, the uncertain parameter ω k , Obey P K and the fuzzy uncertain set represented by P; Π is P K The joint distribution of and P;

[0012] S1.2, Analysis of uncertain data characteristics in highway environment;

[0013] Based on the robust optimization principle, a robust set that can cover the output power of all distributed microgrids is improved, and the distributed robust uncertain data set belonging to a certain probability distribution is constructed in combination with the Wasserstein distance; the above uncertain data set can be described by formula (2) and formula (3):

[0014]

[0015] Where U is a box-type robust uncertain data set, and P is set max , P min are the upper and lower limits of the robust interval of f(x,u) under constraints, represents all probability distributions followed by parameter Ξ.

[0016] Due to the constraints of the confidence interval ε(K,t), the fuzzy uncertain set P presents a K The confidence interval of the box-shaped Wasserstein data set with centered at ε(K, t) and radius ε(K, t) is constructed. The distributed robust uncertain data set constructed under the constraint of this confidence interval contains all possible probability distributions of uncertain parameters.

[0017] S1.3, Dynamic Wasserstein Uncertain Dataset;

[0018] The Wasserstein data set obtained based on step S1.2 is still an independent uncertainty set at time t, and does not have the characteristics of changing with time. By placing the static data set in a rolling time domain environment, a Wasserstein data set with a radius of ε(K, t) that changes dynamically with time is obtained. The mathematical expression of the confidence interval radius ε of this data set is as follows:

[0019]

[0020] In the formula, β is the manually set confidence level of uncertain parameters; D is the sample uniformity coefficient, which can be obtained by solving the optimal probability expectation problem of historical data samples; μ is the average value of sample historical data; ρ is the Euclidean distance between each probability distribution.

[0021] The output power of distributed microgrid is described by constructing Weibull distribution and Beta distribution Wasserstein ball uncertain data sets.

[0022] At the same time, the Wasserstein rectangular data set of electric vehicle charging load is obtained by fitting the fourth-order Gaussian mixture model through the least squares method to realize the uncertain data modeling at the load end;

[0023] S2, "transportation-electricity" coupled ADN network containing SOP equipment;

[0024] Firstly, the highway traffic network and urban microgrids along the highway are modeled separately; secondly, in order to realize the dynamic interaction of urban microgrids, an inter-city microgrid active distribution network (ADN) model with intelligent soft open point (SOP) is proposed; then, the disturbance of highway load on urban microgrids and the geographical distance of urban microgrids are analyzed, and the ADN construction objective function with multiple decision variables is constructed; finally, the NSGA-II algorithm is combined to solve the optimal ADN construction plan;

[0025] S2.1. Highway traffic network model

[0026] The travel distribution model Origin-destination model (OD model) is used to establish the highway model; in addition, the charging piles that bear the charging load of electric vehicles are concentrated in the service area, and the charging load of electric vehicles causes too much disturbance to the power grid and cannot be evenly distributed in the nodes of the urban microgrid; the loads of the remaining highway nodes are conventional loads, and the load power can be evenly distributed in the nodes of the urban microgrid. Therefore, in order to highlight the characteristics of the service area nodes, the service area nodes are modeled in detail; through the description of S2.1, a 25-node highway OD model including 8 service area nodes is finally constructed;

[0027] S2.2. Urban distribution network modeling

[0028] The branch flow model (BFM) is used to model the urban microgrids along the highway. To simplify the calculation, the standard IEEE-33 node model is selected to uniformly model the urban microgrids along the highway.

[0029] Among them, the power balance constraints of the BFM model are as follows:

[0030]

[0031] Where P i , Q i Respectively represent the active power and reactive power values ​​of the i-th node in the power supply network system; G ij , B ij Respectively represent the conductance and susceptance between the i-th node and the j-th node of the microgrid; N represents the total number of microgrid nodes; δ ij is the voltage phase angle difference between the i-th node and the j-th node.

[0032] S2.3. Intercity Microgrid ADN Model with SOP Equipment

[0033] The OD model constructed in S2.1 is combined to extract microgrids along the highway, and an intercity microgrid ADN model with 33 geographical units is constructed, including 3 wind microgrids, 5 photovoltaic microgrids and 25 urban microgrids.

[0034] S2.4. “Transportation-Electricity” ADN Model with SOP

[0035] The construction of the "transportation-electricity" AND model with SOP will be carried out in the following steps. First, in order to ensure the reliability of highway energy supply, the highway nodes are coupled with urban microgrids nearby to form a "transportation-electricity" coupling system; then, appropriate parameters are selected to construct the ADN network benign evaluation system, and the optimal ADN construction objective function is constructed; finally, the NSGA-II algorithm is used to calculate the optimal ADN construction plan.

[0036] S3, Multi-optimization objective collaborative participation second-order optimization model and its adaptive solution algorithm;

[0037] S3.1, Multi-optimization objectives collaboratively participate in the second-order optimization model;

[0038] The structure of the second-order optimization model is as follows: the upper-level main problem calculates the optimal switching scheme of SOP, and combines the global voltage offset of the microgrid to solve the optimal switching state of SOP; the lower-level sub-problem is to calculate the optimal power scheduling scheme of SOP, and the optimal active and reactive power scheduling scheme of SOP is calculated with the minimization of power scheduling cost and the minimization of global voltage offset of microgrid as the optimization objectives;

[0039] S3.2, construction of second-order optimization algorithm and algorithm constraints;

[0040] The PSO algorithm is used to solve the optimal power scheduling scheme, and the "NSGA-II-PSO" two-layer heuristic algorithm is used to construct a second-order multi-objective optimization model to solve the benign ADN network and network power scheduling scheme containing SOP constructed in the rolling time domain; the constraints of the two-layer heuristic optimization algorithm constructed by this method are as follows:

[0041]

[0042]

[0043] Among the above constraints, U bus t They are the upper and lower limits of the voltage of the busbar of the power distribution system; U bus t is the average bus voltage of the microgrid at time t, U grid t They are the upper and lower limits of the grid bus voltage respectively. P grid t and Q grid t are respectively the upper and lower limits of active power and reactive power of the distribution network at time t; grid t , Q grid t are respectively the active power and reactive power generated autonomously by the nth microgrid at time t; P SOP,in,n t , Q SOP,in,n t are respectively the active power and reactive power input by the SOP device to the nth microgrid; similarly, P SOP,out,n t , Q SOP,out,n t Output active power and reactive power for the nth microgrid through the SOP device; P SOP,in t , Q SOP,in t and P SOP,out t , Q SOP,out t They are respectively the upper and lower limits of active power and reactive power that SOP can deliver to the microgrid and the upper and lower limits of active power and reactive power output by SOP. Equations (6) and (7) are the power balance constraints of active power and reactive power; considering that photovoltaic power generation equipment usually does not generate reactive power, the reactive power balance does not consider the photovoltaic power generation factor. is a 0-1 binary variable representing the SOP switch;

[0044] Furthermore, in S1.3 dynamic Wasserstein uncertain data set, the Wasserstein data set formula containing specific probability distribution characteristics is constructed as shown below:

[0045]

[0046]

[0047] In the formula, They are the probability distribution of photovoltaic power generation following Beta distribution; are the probability distribution of wind power generation following the Weibull distribution constraint; π(p 1 ,p 2 ) is the probability distribution p 1 and p 2 The distance measure between them; γ is the distance order, which is an artificially set variable. This parameter is used to describe the integral of the Wasserstein distance difference at the γ order, and represents the measurement accuracy of the Wasserstein distance;

[0048]

[0049] Formula (19) and Formula (20) are expressions of Weibull distribution and Beta distribution respectively. In Formula (19), c is the wind speed parameter; k is the wind speed fluctuation shape parameter. In Formula (20), α and β are the shape parameters of Beta distribution respectively; E max is the photovoltaic output power value under maximum illumination; E is the photovoltaic output power in the actual scene;

[0050] The uncertain data modeling at the load end is as follows:

[0051] First, the uncertain parameter distribution of the load end is simplified into independent normal distributions that are not subordinate to each other, and then the fourth-order Gaussian mixture model (GMM) is shaped by the least squares method to achieve the purpose of constructing load uncertainty data modeling; the formula is as follows:

[0052]

[0053] Where K is the fitting order of GMM, which is an artificially set constant. If the GMM model is fitted by two independent Gaussian models, K = 2; α k is a weight factor used to represent each Gaussian component model N(x,μ k ,∑ k ), which satisfies the constraints given in equation (22); μ k is the location parameter of the kth Gaussian distribution, ∑ k is the dispersion measure parameter of the kth Gaussian distribution;

[0054] Then, the static data set is placed in the rolling time domain to construct a dynamic uncertain data set with rolling optimization capability. Combined with formula (23), the radius of the confidence interval of the uncertain data set ε(K, t) is used to achieve the purpose of constructing a dynamic data set.

[0055]

[0056] ε(K,t) is the data boundary of the uncertain data set P at time t. It is a parameter that can be changed based on manual settings. When time t is set to 0:00~23:59, the uncertain parameter set for the whole day can be obtained.

[0057] Further, S2.4, “transportation-electricity” ADN model with SOP;

[0058] Combined node distance d ij An improved coupling objective function is constructed to avoid the increase in power supply costs and the decrease in energy supply quality caused by incorrect selection of "transportation-electricity" coupling points. The objective function is shown in the following formula:

[0059]

[0060] μ 1 +μ 2 =1 (25)

[0061] In the formula, represents the voltage variance of the jth grid at the i-th moment under the current network coupling scheme; d j represents the geographical location of the j-grid; V k represents the voltage of node k in the power grid under the current SOP construction plan; V ave It represents the average voltage of a single power grid under the current SOP construction plan; T is the time variable; N is the urban microgrid number; μ is the weight coefficient.

[0062] Furthermore, in S3, multiple optimization objectives are collaboratively involved in the second-order optimization model and its adaptive solution algorithm. In the upper-layer SOP switch state calculation model, the minimization of the system global voltage offset under rolling optimization is selected as the objective function to carry out the site selection optimization work. The distribution system voltage offset minimum objective function can be expressed as formula (26):

[0063]

[0064] stt∈[0,24]

[0065] The mathematical expression of the cost objective function in the lower-level power scheduling model is shown in formula (27):

[0066] f 2 =C cons +Cactive +C reactive +C penalty (27)

[0067] In the formula, C cons is the SOP construction cost; C active The cost of dispatching active power for SOP; C reactive is the cost of dispatching reactive power for SOP, C penalty Penalty cost for system voltage excursion;

[0068] C cons =α 1 N 2 +α 2 N (28)

[0069] Where N is the number of SOP construction; α 1 , α 2 Build cost coefficient for SOP;

[0070]

[0071] Where n is the number of SOP construction, n = (1, 2, ... 6); p t ,q t are the active power purchase price and reactive power purchase price at time t respectively; P SOP , Q SOP are the active power and reactive power dispatched by SOP respectively; β 1 , β 2 , β 3 and c 1 、c 2 、c 3 are price constants respectively;

[0072]

[0073] In the formula, is the system voltage variance; ξ 1 , 2 , 3 is the voltage offset penalty parameter.

[0074] Furthermore, in S3.2, the NSGA-II algorithm is used as the upper-level optimization algorithm in the construction of the second-order optimization algorithm and the algorithm constraints to solve the switching strategy of the SOP settings in each time period; when faced with continuous variables and linear constraints such as power scheduling, the PSO algorithm is used to solve the scheduling power value.

[0075] Compared with the prior art, the present invention has the following beneficial effects:

[0076] (1) The "traffic-electricity" coupling network constructed in combination with the specific scenario of the highway can have the overall characteristics of the traffic network OD model, while also having the individual characteristics of an independent BFM. It can effectively incorporate the distributed microgrid and highway demand into the optimized scheduling, and can further improve the safety and economy of the transportation network and microgrid.

[0077] (2) Compared with traditional ADN, the ADN network coupled with SOP switch devices has stronger information interaction and power transmission capabilities between power grids, which enables the redundant power to be effectively utilized, thereby improving the voltage level of the ADN microgrid. This advantage makes the model of the present invention more advantageous in dynamic time domain scheduling, which is specifically reflected in the more stable and uniform electrical indicators of the ADN network containing SOP.

[0078] (3) The data-driven effect of the second-order optimization model "NSGA-II-PSO" constructed by the present invention is good, which is specifically manifested in that the optimization model can flexibly switch the SOP on-off state based on the specific power environment and obtain the optimal power dispatching scheme for the main and sub-problems. Compared with similar optimization models, it not only considers the global electrical stability characteristics of ADN, but also takes into account the economic efficiency of power dispatching. BRIEF DESCRIPTION OF THE DRAWINGS

[0079] Figure 1 This is the map of the "Taohuaping-Dingyuan" section of the G30 "Lianhuo Expressway";

[0080] Figure 2 It is the power flow diagram of the "transportation-electricity" coupling network;

[0081] Figure 3 It is the "transportation-electricity" coupling network diagram;

[0082] Figure 4 The voltage amplitude curve of the ADN independent power grid under different simulation scenarios;

[0083] Figure 5 It is the SOP operating diagram and SOP transmission power value curve. DETAILED DESCRIPTION

[0084] In order to make the technical means, creative features, objectives and effects achieved by the present invention easy to understand, the present invention is further explained below in conjunction with specific implementation methods.

[0085] The present invention provides an intelligent soft switching method for operation and scheduling of an active distribution network of a highway. The technology is divided into the following steps:

[0086] S1. Highway multivariate uncertain data model based on Wasserstein distance;

[0087] S1.1, Wasserstein distance definition and its application characteristics;

[0088] Wasserstein distance, also known as Earth-Mover's Distance, is a measure of the similarity between two probability distributions. Unlike traditional statistical distances, such as KL divergence and JS divergence, Wasserstein distance can reflect the distance between two probability distributions even when there is no overlap. The calculation of this distance involves the optimal transmission problem, that is, how to transform one probability distribution into another probability distribution under a certain cost function so that the total cost is minimized.

[0089] The definition of Wasserstein distance involves all possible joint distributions between two probability distributions, and the final distance value is obtained by taking the lower bound of the expected value of the distance between the samples under these joint distributions. Specifically, for two probability distributions P and P K , its Wasserstein distance is defined as:

[0090]

[0091] In the formula, the uncertain parameter ω k , Obey P K and the fuzzy uncertain set represented by P; Π is P K The joint distribution of and P.

[0092] Compared with the KL divergence model and the JS divergence model, which require specific data as the basis for evaluating the similarity of probability distribution, the Wasserstein distance does not require specific data as the information carrier of probability distribution. When the uncertain data set constructed based on the Wasserstein distance faces the highway scenario, due to the high difficulty of predicting the EV charging load on the highway and the large differences in the electrical characteristics of the distributed microgrid due to the long geographical span, constructing a fuzzy uncertain data set consisting only of upper and lower intervals is more conducive to simplifying the data model of the ADN system.

[0093] S1.2, Analysis of uncertain data characteristics in highway environment;

[0094] ADN has both conventional loads and new energy generation, which are random elements, and adjustable resources such as energy storage and flexible loads. It may be flexible as a whole, or it may be random like a traditional distribution network. Before describing the operation domain of the gateway power, it should be determined whether it has the ability to provide such an operation domain.

[0095] The distributed microgrids in Gansu are mainly composed of distributed photovoltaic microgrids and distributed wind power microgrids. Their power characteristics show strong volatility and randomness. It is unrealistic to use a certain probability distribution or power curve to refer to the output power of the distributed microgrid. Therefore, based on the robust optimization principle, the present invention proposes a robust set that can include the output power of all distributed microgrids, and combines the Wasserstein distance to construct a distributed robust uncertain data set subject to a certain probability distribution. The above uncertain data set can be described by formula (2) and formula (3):

[0096]

[0097] Where U is a box-type robust uncertain data set, and P is set max , P min They are respectively the upper and lower limits of the robust interval of f(x,u) under the constraints. represents all probability distributions followed by the z parameter Ξ. It is worth noting that due to the constraints of the confidence interval ε(K,t), the fuzzy uncertain set P presents a P K The confidence interval of the box-shaped Wasserstein data set with centered at ε(K, t) and radius ε(K, t) is constructed. The distributed robust uncertain data set constructed under the constraint of this confidence interval contains all possible probability distributions of uncertain parameters.

[0098] S1.3, Dynamic Wasserstein Uncertain Dataset;

[0099] The Wasserstein data set obtained based on step S1.2 is still an independent uncertain set at time t, and does not have the characteristics of changing over time. In order to ensure the real-time performance of highway ADN power scheduling, it is necessary to construct an uncertain data set that can reflect the dynamic changes in the power of distributed microgrids. Therefore, the present invention places the static data set in a rolling time domain environment to obtain a Wasserstein data set with a radius of ε(K, t) that changes dynamically over time. The mathematical expression of the confidence interval radius ε of the data set is as follows.

[0100]

[0101] In the formula, β is the manually set confidence level of uncertain parameters; D is the sample uniformity coefficient, which can be obtained by solving the optimal probability expectation problem of historical data samples; μ is the average value of sample historical data.

[0102] Considering that the distributed microgrids along the highways are mainly equipped with wind power generation and photovoltaic power generation equipment, and the wind power and photovoltaic power obey the Weibull distribution and Beta distribution respectively, the output power of the distributed microgrid is described by constructing Weibull distribution and Beta distribution Wasserstein ball uncertain data sets.

[0103] The uncertainty of the highway load end mainly comes from the electric vehicle charging load, which is attributed to the fact that the electric vehicle charging load is difficult to be described by a certain probability distribution. To deal with the above problem, the present invention obtains the Wasserstein rectangular data set of the electric vehicle charging load by fitting the fourth-order Gaussian mixture model by the least squares method, and realizes the uncertain data modeling of the load end.

[0104] S2, "transportation-electricity" coupled ADN network containing SOP equipment;

[0105] Building a dynamic "transportation-electricity" coupling system is a prerequisite for solving the optimal power dispatching strategy, but the huge differences between the transportation network and the power supply network make it difficult to build a dynamic coupling system. From a spatial perspective, highway electricity is usually provided by cities along the highway, and the radiation range of urban microgrids is much smaller than the distance between highway nodes; from the perspective of electrical characteristics, the electrical characteristics of each node vary greatly; from the perspective of information type, the transportation network collects dynamic information of individual decisions of electric vehicles, while the power grid is static information of power flow. In summary, building a coupling system is essentially a dynamic interaction of "dynamic-static" multi-information heterogeneous networks.

[0106] To solve the above problems, the present invention first models the highway traffic network and the urban microgrids along the highway respectively. Secondly, in order to realize the dynamic interaction of urban microgrids, an inter-city microgrid active distribution network (ADN) model with intelligent soft open point (SOP) is proposed. Then, the disturbance of highway load on urban microgrids and the geographical distance of urban microgrids are analyzed, and the ADN construction objective function with multiple decision variables is constructed. Finally, the NSGA-II algorithm is combined to solve the optimal ADN construction plan.

[0107] S2.1. Highway traffic network model

[0108] In order to describe the randomness of urban electric vehicles, the academic community usually combines the user equilibrium model (EU), flow capturing location model (FCLM) and flow fueling location model (FRLM) to build improved models when constructing urban road network models. However, the randomness of highway vehicles in terms of route selection and charging decisions is weak, and the highway model can be described using geographic data sets.

[0109] In view of the above characteristics, the present invention adopts the travel distribution model (Origin-destination model, O-Dmodel) to establish a highway model. In addition, the charging piles that bear the charging load of electric vehicles are concentrated in the service area, and the charging load of electric vehicles causes too much disturbance to the power grid and cannot be evenly distributed in the nodes of the urban microgrid; while the loads of the remaining highway nodes are conventional loads, and the load power can be evenly distributed in the nodes of the urban microgrid. Therefore, in order to highlight the characteristics of the service area nodes, the present invention focuses on modeling the service area nodes. Through the description of S2.1, a 25-node highway OD model containing 8 service area nodes was finally constructed.

[0110] S2.2. Urban distribution network modeling

[0111] Existing power system analysis is based on power flow calculation. On the one hand, it is difficult to extract data from traditional AC power flow models and cannot interact with the optimal power flow objective function. On the other hand, the Distflow DC power flow model not only requires the convexity of the optimal power flow function, but also artificially ignores the impact of reactive power, which limits its application scope.

[0112] After comprehensive comparison, the present invention uses the branch flow model (BFM) to model the urban microgrid along the highway. To simplify the calculation, the present invention uses the standard IEEE-33 node model to uniformly model the urban microgrid along the highway. In this system, the voltage level is 10.5kV, the total active power is 3715kW, and the total reactive power is 2300kVAR.

[0113] The power balance constraints of the BFM model are as follows.

[0114]

[0115] Where P i , Q i Respectively represent the active power and reactive power values ​​of the i-th node in the power supply network system; G ij , B ijThey represent the conductance and susceptance between the i-th node and the j-th node of the microgrid respectively; N represents the total number of microgrid nodes.

[0116] S2.3. Intercity Microgrid ADN Model with SOP Equipment

[0117] In the highway power supply network, since the power supply capacity of a single city microgrid cannot radiate to each service area of ​​the highway, the highway power supply network is composed of a series of urban distribution networks along the line. SOP equipment has important value in realizing information communication and power exchange between urban microgrids. Using SOP to connect urban microgrids along the highway and build a highway power supply network with a long spatial span has important practical significance.

[0118] The ADN model is a dynamic grid model that can use electrical equipment to change the interconnection state of grid nodes. This feature enables ADN to reorganize the node connection state, disperse the extreme power of some nodes to appropriate nodes, and call on the regulation capabilities of each interconnected node to offset extreme power shocks and improve system resilience. SOP assumes the function of node interconnection equipment, and building an ADN model with SOP provides a new method for building intercity microgrids on highways.

[0119] The present invention extracts microgrids along the highway in combination with the OD model constructed in S2.1, and constructs an intercity microgrid ADN model containing 33 geographical units, including 3 wind microgrids, 5 photovoltaic microgrids and 25 urban microgrids.

[0120] S2.4. “Transportation-Electricity” ADN Model with SOP

[0121] Furthermore, it is necessary to couple the independent transportation network, independent microgrid and SOP equipment to construct a "transportation-electricity" ADN model containing SOP. The "transportation-electricity" coupling process can be regarded as the process of injecting load into the urban power grid through charging piles on the highway. Considering that charging piles are usually concentrated in service areas, the service area nodes are mainly considered when constructing the "transportation-electricity" coupling network.

[0122] Combined with the above content, the present invention will construct the "transportation-electricity" AND model containing SOP from the following steps. First, in order to ensure the reliability of highway energy supply, the highway node selects the nearest urban microgrid coupling to form a "transportation-electricity" coupling system. Then, select appropriate parameters to construct the ADN network benign evaluation system, and construct the optimal ADN construction objective function. Finally, the NSGA-II algorithm is used to calculate the optimal ADN construction plan.

[0123] S3, Multi-optimization objective collaborative participation second-order optimization model and its adaptive solution algorithm;

[0124] S3.1, Multi-optimization objectives collaboratively participate in the second-order optimization model;

[0125] The second-order optimization model is a mathematical model that places multiple objectives at different optimization levels for collaborative optimization. Compared with the traditional model that sets weights to couple multiple objectives into a single objective and then optimizes, the second-order optimization model can solve the main problem while taking into account the optimality of the sub-problem solution, avoiding falling into the local optimal solution. This feature gives the second-order optimization model an outstanding advantage in solving the main and sub-objective game problem.

[0126] Setting the dynamic SOP switching scheme based on rolling power flow parameters is the primary task for the stable operation of the ADN system. Therefore, the second-order optimization model structure proposed in this invention is as follows: the upper-level main problem calculates the optimal SOP switching scheme, and the optimal switching state of SOP is solved in combination with the global voltage offset of the microgrid. The lower-level sub-problem is to calculate the optimal power scheduling scheme of SOP, and the optimal active and reactive power scheduling scheme of SOP is calculated with the optimization objectives of minimizing the power scheduling cost and minimizing the global voltage offset of the microgrid.

[0127] S3.2, construction of second-order optimization algorithm and algorithm constraints;

[0128] It can be seen from the data model of electric vehicle charging load and wind-solar microgrid power generation constructed in the previous text that the "ADN reconstruction-power scheduling" second-order optimization model constructed by the present invention needs to process the continuous parameters of source and load access on the one hand; on the other hand, it needs to process the binary nonlinear parameters that characterize the SOP off state. Although the mixed integer linear programming algorithm (MILP) developed based on Cplex and Gurobi linear solvers has been widely proven to have excellent performance in solving linear convex problems under continuous constraints, it is still difficult to process 0-1 binary nonlinear variables in the rolling time domain.

[0129] On the one hand, the artificial genetic factors of the NSGA-II genetic algorithm can better carry non-continuous binary variables, and on the other hand, the cross-variation characteristics of the genetic factors enable the algorithm to adapt to nonlinear environments. Based on the above considerations, the present invention selects the NSGA-II algorithm to solve the SOP closing / opening scheme. Particle swarm optimization (PSO) is a heuristic algorithm that simulates bird flocks foraging. Compared with traditional optimization algorithms, it has the advantages of fast convergence speed and weak convexity requirements. The present invention uses the PSO algorithm to solve the optimal power scheduling scheme. Therefore, in combination with the two-level programming problem in the problem description, the present invention uses the "NSGA-II-PSO" two-level heuristic algorithm to construct a second-order multi-objective optimization model to solve the benign ADN network containing SOP constructed in the rolling time domain and the network power scheduling scheme. The constraints of the two-level heuristic optimization algorithm constructed by the present invention are as follows.

[0130]

[0131]

[0132] Among the above constraints, U bus t They are the upper and lower limits of the voltage of the busbar of the power distribution system; U bus t is the average bus voltage of the microgrid at time t, U grid t They are the upper and lower limits of the grid bus voltage respectively. P grid t and Q grid t are respectively the upper and lower limits of active power and reactive power of the distribution network at time t; grid t , Q grid t are respectively the active power and reactive power generated autonomously by the nth microgrid at time t. SOP,in,n t , Q SOP,in,n t The active power and reactive power are input to the nth microgrid through the SOP device. SOP,out,n t , Q SOP,out,n t Output active power and reactive power for the nth microgrid through the SOP device. P SOP,in t , Q SOP,in t and P SOP,out t , Q SOP,out tThey are respectively the upper and lower limits of active power and reactive power that SOP can deliver to the microgrid and the upper and lower limits of active power and reactive power output by SOP. Equations (6) and (7) are the power balance constraints of active power and reactive power; considering that photovoltaic power generation equipment usually does not generate reactive power, the reactive power balance does not consider the photovoltaic power generation factor. i t is a 0-1 binary variable representing the SOP switch.

[0133] S1.3 Dynamic Wasserstein uncertain data set;

[0134] The present invention realizes data modeling and abstract mathematical expression of uncertain data by constructing a Wasserstein data set that can include any possible power probability distribution of distributed microgrids. In addition, by artificially changing the confidence interval radius ε of the Wasserstein data set, the present invention successfully realizes the flexible control of the boundary of the "source-load" box-type uncertain data set, so that uncertain data can flexibly approach or move away from extreme scenarios, and endows the power dispatching flexibility and real-time performance of the highway ADN system from the data-driven level.

[0135] As described in S1.2 of the present invention, the uncertain data of the highway power system at the power supply end mainly comes from the wind power generation equipment and photovoltaic power generation equipment of the distributed microgrid; for the load end, the uncertain data comes from the EV (Electrical vehicle.EV) charging load.

[0136] The random sampling method has been widely used to obtain the uncertain parameters of concrete power output simulation, but the probability distribution has a stronger advantage in describing the characteristics of uncertain parameters due to its strong information storage capacity. At present, the output power of distributed microgrids is usually represented by continuous probability distribution, while in the dispatching model, discrete distribution is generally used to replace continuous distribution for simplification.

[0137] By consulting existing research results, it is known that the photovoltaic power generation in distributed microgrids follows the Beta distribution constraint, while the wind power generation follows the Weibull distribution constraint. Combined with the definition of Wasserstein distance shown in formula (1) in S1.1, we can further obtain a box-type Wasserstein data set that combines the characteristics of Beta distribution and Weibull distribution. The formula of the constructed Wasserstein data set containing specific probability distribution characteristics is shown as follows:

[0138]

[0139] In the formula, They are the probability distribution of photovoltaic power generation following Beta distribution; are the probability distribution of wind power generation following the Weibull distribution constraint; π(p 1 ,p 2 ) is the probability distribution p 1 and p 2 The distance measure between them; γ is the distance order, which is an artificially set variable. This parameter is used to describe the integral of the Wasserstein distance difference at the γ order, and characterizes the measurement accuracy of the Wasserstein distance.

[0140]

[0141] Formula (19) and Formula (20) are expressions of Weibull distribution and Beta distribution respectively. In Formula (19), c is the wind speed parameter; k is the wind speed fluctuation shape parameter. In Formula (20), α and β are the shape parameters of Beta distribution respectively; E max is the photovoltaic output power value under maximum illumination; E is the photovoltaic output power in the actual scene.

[0142] Taking into account the unpredictability of EV charging load on highways, in order to characterize the uncertain parameters at the load end, the present invention first simplifies the distribution of uncertain parameters at the load end into independent normal distributions that are not subordinate to each other, and then shapes the fourth-order Gaussian mixture model (GMM) through the least squares method to achieve the purpose of constructing load uncertainty data modeling.

[0143]

[0144] Where K is the fitting order of GMM, which is an artificially set constant. If the GMM model is fitted by two independent Gaussian models, K = 2; α k is a weight factor used to represent each Gaussian component model N(x,μ k ,∑ k ), which satisfies the constraints given in equation (22); μ k is the location parameter of the kth Gaussian distribution, ∑ k It is k Dispersion measure parameter for the Gaussian distribution.

[0145] The previous content constructs the Wasserstein data set in the static time domain. The present invention further places the static data set in the rolling time domain to construct a dynamic uncertain data set with rolling optimization capability. The present invention combines formula (23) and uses the radius ε(K, t) of the confidence interval of the uncertain data set to achieve the purpose of constructing a dynamic data set.

[0146]

[0147] ε(K,t) is the data boundary of the uncertain data set P at time t, and is a parameter that can be changed based on manual settings. When time t is set to 0:00~23:59, the uncertain parameter set for the whole day can be obtained.

[0148] Based on the Wasserstein data set constructed in this part, the present invention realizes the characterization of the "source-load" uncertain parameters of the highway power supply system. In addition, through the confidence interval radius ε(K,t) that can be changed in real time, the present invention realizes the flexible change of the electrical simulation environment, improves the data-driven ability of the simulation system, and also builds an environment for the robust optimization of the ADN system.

[0149] S2.4, “Transportation-Electricity” ADN model with SOP;

[0150] In steps S2.1 and S2.2, a long-distance OD traffic model that meets the characteristics of a highway and a chain electrical network that is different from a traditional urban network are constructed. Furthermore, the present invention briefly introduces the basic components and spatial characteristics of the ADN containing SOP equipment in step S2.3. In the subsequent step S2.4, the independently operated traffic network and power grid are coupled to form a "traffic-electrical" coupling network that can form information interaction, energy exchange, and equipment complementarity.

[0151] By analyzing the process of injecting EV charging loads from the highway traffic network service area into the independent power grid along the line and the power flow impact brought by the EV load grid connection, it is found that the fluctuation impact of the uncertain power after grid connection will eventually be reflected in the ADN system voltage. In order to ensure that the obtained "traffic-electrical" coupling system has basic anti-disturbance capabilities, the present invention refers to the idea of ​​the FCLM model that the traffic grid connection has the least impact on the power grid, and selects the ADN intraday global voltage deviation as the evaluation standard for the "traffic-electrical" network coupling quality. In addition, considering that the energy supply coverage of independent urban power grids in the context of highways is limited, if the "traffic-electrical" coupling point is selected incorrectly, it will lead to an increase in the cost of electricity supply and a decrease in the quality of energy supply. Therefore, the present invention combines the node distance d ij Construct an improved coupling objective function, which is shown in the following formula:

[0152]

[0153] μ 1 +μ 2 =1 (25)

[0154] In the formula, represents the voltage variance of the jth grid at the i-th moment under the current network coupling scheme; d j represents the geographical location of the j-grid; V k represents the voltage of node k in the power grid under the current SOP construction plan; V aveIt represents the average voltage of a single power grid under the current SOP construction plan; T is the time variable; N is the urban microgrid number; μ is the weight coefficient.

[0155] To ensure the universality of the resulting coupling network, the present invention constructs a coupling network under typical wind and solar output power and typical EV charging load environments. And to ensure that the calculation process always follows the constraints of the objective function, the present invention uses the NSGA-II algorithm with good convergence performance and high calculation speed to calculate the coupling results.

[0156] S3, Multi-optimization objective collaborative participation second-order optimization model and its adaptive solution algorithm;

[0157] In S3.1, the present invention discusses the feasibility and adaptability of constructing a second-order optimization model for a "transportation-electricity" coupling network, and proves that the second-order optimization algorithm can carry multi-objective constraints and carry out multi-objective collaborative optimization. In view of the good optimization effect of the second-order optimization model, the present invention carries out optimization with robustness and economy as optimization goals.

[0158] The upper SOP switch state calculation model: S2.2 content proves that the FCLM model has excellent performance in exerting the power supply service capacity of microgrids and ensuring the stability of highway power grids. Therefore, the present invention continues to select the minimization of the global voltage offset of the system under rolling optimization as the objective function to carry out site selection optimization. The distribution system voltage offset minimum objective function can be shown by formula (26).

[0159]

[0160] stt∈[0,24]

[0161] Lower-layer power dispatch model: Although the ADN system equipped with SOP can reduce extreme load shocks and improve the flexibility of the power grid in the face of sudden load changes, the SOP construction cost and SOP dispatch cost still limit the application scope of ADN. Carrying out cost optimization to ensure the economic efficiency of the ADN system is another essential task.

[0162] The cost objective function defined in the lower optimization model of the present invention includes SOP construction cost and SOP power dispatch cost. At the same time, the grid stability should also be taken into consideration, and a penalty cost function is constructed in combination with the microgrid voltage offset. The mathematical expression of the cost objective function that integrates multiple factors is shown in formula (27).

[0163] f 2 =C cons +C active +C reactive +C penalty (27)

[0164] In the formula, Ccons is the SOP construction cost; C power The cost of dispatching active power for SOP; C reactive The cost of dispatching reactive power for SOP. penalty Penalty cost for system voltage excursion.

[0165] C cons =α 1 N 2 +α 2 N (28)

[0166] Where N is the number of SOP construction; α 1 , α 2 Construction cost coefficient for SOP.

[0167]

[0168] Where n is the number of SOP construction, n = (1, 2, ... 6); p t ,q t are the active power purchase price and reactive power purchase price at time t respectively; P SOP , Q SOP are the active power and reactive power dispatched by SOP respectively; β 1 , β 2 , β 3 and c 1 、c 2 、c 3 are price constants respectively.

[0169]

[0170] In the formula, is the system voltage variance; ξ 1 , 2 , 3 is the voltage offset penalty parameter.

[0171] S3.2, construction of second-order optimization algorithm and algorithm constraints;

[0172] Different from the widely used first-order optimization model, the present invention constructs a two-layer multi-objective collaborative optimization model in S3.1 that adapts to the data-driven capabilities of uncertain data sets. However, considering that the main application scenario of the widely used linear programming is the first-order optimization model, in order to distinguish the second-order optimization model and improve the solving efficiency of the second-order optimization problem, the present invention constructs a two-layer heuristic algorithm based on the heuristic algorithm.

[0173] In view of the excellent adaptability of the NSGA-II algorithm when carrying 0-1 binary variables and the convergence when solving the "traffic-electricity" coupling network, the present invention uses the NSGA-II algorithm as the upper optimization algorithm to solve the switching strategy of the SOP setting in each time period. When facing continuous variables and linear constraints such as power scheduling, the present invention uses the PSO algorithm to solve the scheduling power value.

[0174] Example:

[0175] In order to verify the feasibility and effectiveness of the practice of the present invention, the present invention takes the "Taohuaping-Dingyuan" section of the G30 "Lianhuo Expressway" as the research object. First, the present invention uses ArcGIS software to extract the spatial geographic information of the expressway, arranges the power grid and traffic nodes along the expressway in the OD model, and constructs independent traffic models and power grid models. The object expressway map obtained by ArcGIS software is as follows Figure 1 shown.

[0176] For the power grids along the highway, the present invention divides the independent power grids along the highway into three types: county and town power grids, urban power grids, and distributed microgrids according to the power grid capacity and power generation scale; to simplify the power grid model, the present invention uses the IEEE-33 node power flow model to uniformly model various power grids. In terms of power flow calculation, the present invention uses the branch flow model (BFM) with outstanding parameter operability advantages to calculate electrical power flow parameters.

[0177] Then, as described in step S2.4, the present invention constructs a "transportation-electricity" coupling model based on the improved FCLM model shown in formula (12). The resulting coupling system Figure 3 shown.

[0178] Subsequently, based on Figure 3 The "transportation-electricity" coupling system shown in the figure carries out multi-objective collaborative optimization. The steps of the present invention for carrying out optimization scheduling are as follows:

[0179]

[0180]

[0181]

[0182] Finally, in order to verify the effectiveness of the method proposed in this invention, the present invention calculates the peak load regulation scheme of the method proposed in this invention, and at the same time conducts comparative experiments on some current mainstream methods in academia. The SOP switching scheme and dispatching power numerical power generation curve and peak load regulation cost obtained by various methods are as follows: Figure 4In addition, placing the proposed model in different uncertain data environments can verify the robustness of the proposed system, and the data are shown in Table 1.

[0183] As can be seen from Table 1, the method proposed in the present invention can carry out peak shaving work in an operating environment with large peak-to-valley difference in a short period of time, and ensure the stability of electrical parameters of the grid-connected ADN system. However, the peak shaving of the present invention will have a small negative impact on the system frequency.

[0184] Table 1 Optimization parameters of ADN system including SOP

[0185]

[0186] Table 2 is a comparative data table of similar methods. The data comparison shown in Table 2 further proves the excellent performance of the model proposed in the present invention in integrating electrical equipment along the highway. Compared with the traditional ADN, the ADN network coupled with SOP switchgear has stronger information interaction and power transmission capabilities between power grids, which enables the redundant power to be effectively utilized, thereby improving the voltage level of the ADN microgrid. This advantage makes the model of the present invention more advantageous in dynamic time domain scheduling, which is specifically reflected in the fact that the electrical indicators of the ADN network containing SOP are more stable and uniform.

[0187] Table 2 Data comparison of similar methods

[0188]

[0189] It will be apparent to those skilled in the art that the invention is not limited to the details of the exemplary embodiments described above and that the invention can be implemented in other specific forms without departing from the spirit or essential features of the invention. Therefore, the embodiments should be considered exemplary and non-limiting in all respects, and the scope of the invention is defined by the appended claims rather than the foregoing description, and it is intended that all variations falling within the meaning and range of equivalent elements of the claims be included in the invention. Any reference numeral in a claim should not be considered as limiting the claim to which it relates.

[0190] In addition, it should be understood that although the present specification is described according to implementation modes, not every implementation mode contains only one independent technical solution. This description of the specification is only for the sake of clarity. Those skilled in the art should regard the specification as a whole. The technical solutions in each embodiment may also be appropriately combined to form other implementation modes that can be understood by those skilled in the art.

Claims

1. An intelligent soft switching method for operation and dispatching of active distribution networks on expressways, used for power dispatching on electrified expressways, characterized by: The following steps are involved: S1. Highway multivariate uncertain data model based on Wasserstein distance; S1.1, Wasserstein distance definition and its application characteristics; The definition of Wasserstein distance involves all possible joint distributions between two probability distributions, and the final distance value is obtained by taking the lower bound of the expected value of the distance between the samples under these joint distributions. Specifically, for two probability distributions P and P K , its Wasserstein distance is defined as: In the formula, the uncertain parameter ω k , Obey P K and the fuzzy uncertain set represented by P; Π is P K The joint distribution of and P; S1.2, Analysis of uncertain data characteristics in highway environment; Based on the robust optimization principle, a robust set that can cover the output power of all distributed microgrids is improved, and the distributed robust uncertain data set belonging to a certain probability distribution is constructed in combination with the Wasserstein distance; the above uncertain data set can be described by formula (2) and formula (3): Where U is a box-type robust uncertain data set, and P is set max , P min are the upper and lower limits of the robust interval of f(x,u) under constraints, represents all probability distributions followed by parameter Ξ; Due to the constraints of the confidence interval ε(K,t), the fuzzy uncertain set P presents a K The confidence interval of the box-shaped Wasserstein data set with centered at ε(K,t) and radius ε(K,t). The distributed robust uncertain data set constructed under the constraint of this confidence interval contains all possible probability distributions of uncertain parameters. S1.3, Dynamic Wasserstein Uncertain Dataset; The Wasserstein data set obtained based on step S1.2 is still an independent uncertainty set at time t, and does not have the characteristics of changing with time. By placing the static data set in a rolling time domain environment, a Wasserstein data set with a radius of ε(K, t) that changes dynamically with time is obtained. The mathematical expression of the confidence interval radius ε of this data set is as follows: In the formula, β is the manually set confidence level of uncertain parameters; D is the sample uniformity coefficient, which can be obtained by solving the optimal probability expectation problem of historical data samples; μ is the average value of sample historical data; ρ is the Euclidean distance between each probability distribution; Further refined modeling of distributed microgrid output data is carried out; considering that the output power of wind power microgrid and photovoltaic power microgrid follows Weibull distribution and Beta distribution respectively, the corresponding probability distribution is used to construct Wasserstein sphere uncertain data set during data modeling; For the charging load of electric vehicles, the least squares method is used to fit the fourth-order Gaussian mixture model to obtain the corresponding Wasserstein data set to achieve uncertain data modeling at the load end; S2, "transportation-electricity" coupled ADN network containing SOP equipment; Firstly, the highway traffic network and urban microgrids along the highway are modeled separately; secondly, in order to realize the dynamic interaction of urban microgrids, an inter-city microgrid active distribution network (ADN) model with intelligent soft open point (SOP) is proposed; then, the disturbance of highway load on urban microgrids and the geographical distance of urban microgrids are analyzed, and the ADN construction objective function with multiple decision variables is constructed; finally, the NSGA-II algorithm is combined to solve the optimal ADN construction plan; S2.

1. Highway traffic network model The travel distribution model Origin-destination model (OD model) is used to establish the highway model; in addition, the charging piles that bear the charging load of electric vehicles are concentrated in the service area, and the charging load of electric vehicles causes too much disturbance to the power grid and cannot be evenly distributed in the nodes of the urban microgrid; the loads of the remaining highway nodes are conventional loads, and the load power can be evenly distributed in the nodes of the urban microgrid. Therefore, in order to highlight the characteristics of the service area nodes, the service area nodes are modeled in detail; through the description of S2.1, a 25-node highway OD model including 8 service area nodes is finally constructed; S2.

2. Urban distribution network modeling The branch flow model (BFM) is used to model the urban microgrids along the highway. To simplify the calculation, the standard IEEE-33 node model is selected to uniformly model the urban microgrids along the highway. Among them, the power balance constraints of the BFM model are as follows: Where P i , Q i Respectively represent the active power and reactive power values ​​of the i-th node in the power supply network system; G ij , B ij Respectively represent the conductance and susceptance between the i-th node and the j-th node of the microgrid; N represents the total number of microgrid nodes; δ ij is the voltage phase angle difference between the i-th node and the j-th node; S2.

3. Intercity Microgrid ADN Model with SOP Equipment Combined with the OD model constructed in S2.1, the microgrids along the highway are extracted to build an intercity microgrid ADN model with 33 geographical units, including 3 wind microgrids, 5 photovoltaic microgrids and 25 urban microgrids; S2.4, "Transportation-Electricity" ADN Model with SOP The construction of the "transportation-electricity" AND model with SOP will be carried out from the following steps. First, in order to ensure the reliability of highway energy supply, the highway nodes are coupled with urban microgrids nearby to form a "transportation-electricity" coupling system; then, appropriate parameters are selected to construct the ADN network benign evaluation system, and the optimal ADN construction objective function is constructed; finally, the NSGA-II algorithm is used to calculate the optimal ADN construction plan; S3, Multi-optimization objective collaborative participation second-order optimization model and its adaptive solution algorithm; S3.1, Multi-optimization objectives collaboratively participate in the second-order optimization model; The structure of the second-order optimization model is as follows: the upper master problem calculates the optimal switching scheme of the SOP, and the optimal switching state of the SOP is solved in combination with the global voltage offset of the microgrid; The lower-level sub-problem is to calculate the optimal power dispatching scheme of SOP, and to calculate the optimal active and reactive power dispatching scheme of SOP with the optimization objectives of minimizing the power dispatching cost and minimizing the global voltage deviation of the microgrid; S3.2, construction of second-order optimization algorithm and algorithm constraints; The PSO algorithm is used to solve the optimal power scheduling scheme, and the "NSGA-II-PSO" two-layer heuristic algorithm is used to construct a second-order multi-objective optimization model to solve the benign ADN network and network power scheduling scheme containing SOP constructed in the rolling time domain; the constraints of the two-layer heuristic optimization algorithm constructed by this method are as follows: Among the above constraints, U bust They are the upper and lower limits of the voltage of the busbar of the power distribution system; U bust is the average bus voltage of the microgrid at time t, U gridt are the upper and lower limits of the grid bus voltage, P gridt and Q gridt are respectively the upper and lower limits of active power and reactive power of the distribution network at time t; gridt , Q gridt are respectively the active power and reactive power generated autonomously by the nth microgrid at time t; P SOP,in,n t , Q SOP,in,n t are respectively the active power and reactive power input by the SOP device to the nth microgrid; similarly, P SOP,out,n t , Q SOP,out,n t Output active power and reactive power for the nth microgrid through the SOP device; P SOP,in t , Q SOP,in t and P SOP,out t , Q SOP,out t are the upper and lower limits of active power and reactive power that SOP can deliver to the microgrid and the upper and lower limits of active power and reactive power output by SOP, respectively. Equations (6) and (7) are the power balance constraints of active power and reactive power. Considering that photovoltaic power generation equipment usually does not generate reactive power, the reactive power balance does not consider the photovoltaic power generation factor. is a 0-1 binary variable representing the switch state of the SOP.

2. The intelligent soft switching method for operation and dispatching of active distribution network of highway according to claim 1, characterized in that: In S1.3 dynamic Wasserstein uncertain data set, the Wasserstein data set formula containing specific probability distribution characteristics is constructed as shown below: In the formula, They are the probability distribution of photovoltaic power generation following Beta distribution; are the probability distributions of wind power generation that follow the Weibull distribution constraints; π(p1, p2) is the distance measure between the probability distributions p1 and p2; γ is the order of the measured distance, which is an artificially set variable. This parameter is used to describe the integral of the measured Wasserstein distance difference at the γ order, and represents the measurement accuracy of the Wasserstein distance; Formula (19) and formula (20) are expressions of Weibull distribution and Beta distribution respectively. In formula (19), c is the wind speed parameter; k is the wind speed fluctuation shape parameter; in formula (20), Γ is the description formula of Beta distribution; α and β are the shape parameters of Beta distribution respectively; E max is the photovoltaic output power value under maximum illumination; E is the photovoltaic output power in the actual scene; The uncertain data modeling at the load end is as follows: First, the uncertain parameter distribution of the load end is simplified into independent normal distributions that are not subordinate to each other, and then the fourth-order Gaussian mixture model (GMM) is shaped by the least squares method to achieve the purpose of constructing load uncertainty data modeling; the formula is as follows: Where K is the fitting order of GMM, which is an artificially set constant. If the GMM model is fitted by two independent Gaussian models, K = 2; α k is the weight factor used to represent each Gaussian component model N(x,μ k ,∑ k ), which satisfies the constraints given in equation (22); μ k is the location parameter of the kth Gaussian distribution, ∑ k is the dispersion measure parameter of the kth Gaussian distribution; Then, the static data set is placed in the rolling time domain to construct a dynamic uncertain data set with rolling optimization capability. Combined with formula (23), the radius of the confidence interval of the uncertain data set ε(K, t) is used to achieve the purpose of constructing a dynamic data set. ε(K,t) is the data boundary of the uncertain data set P at time t. It is a parameter that can be changed based on manual settings. When time t is set to 0:00~23:59, the uncertain parameter set for the whole day can be obtained.

3. The intelligent soft switching method for operation and dispatching of active distribution network of highway according to claim 1, characterized in that: S2.4, "Transportation-Electricity" ADN model with SOP; Combined node distance d ij An improved coupling objective function is constructed to avoid the increase in power supply costs and the decrease in energy supply quality caused by incorrect selection of "transportation-electricity" coupling points. The objective function is shown in the following formula: μ1+μ2=1 (25) In the formula, represents the voltage variance of the jth grid at the i-th moment under the current network coupling scheme; d j represents the geographical location of the j-grid; V k represents the voltage of node k in the power grid under the current SOP construction plan; V ave It represents the average voltage of a single power grid under the current SOP construction plan; T is the time variable; N is the urban microgrid number; μ is the weight coefficient.

4. The intelligent soft switching method for operation and dispatching of active distribution network of highway according to claim 1, characterized in that: S3. In the second-order optimization model and its adaptive solution algorithm, the minimization of the global voltage offset of the system under rolling optimization is selected as the objective function in the upper-level SOP switch state calculation model to carry out the site selection optimization work. The minimum objective function of the voltage offset of the distribution system can be expressed as formula (26): The mathematical expression of the cost objective function in the lower-level power scheduling model is shown in formula (27): min F2=C cons +C active +C reactive +C penalty (27) In the formula, C cons is the SOP construction cost; C active The cost of dispatching active power for SOP; C reactive is the cost of dispatching reactive power for SOP, C penalty Penalty cost for system voltage excursion; C cons =α1N 2 +α2N (28) Where N is the number of SOP construction; α1 and α2 are SOP construction cost coefficients; Where n is the number of SOP construction, n = (1, 2, ... 6); p t ,q t are the active power purchase price and reactive power purchase price at time t respectively; P SOP , Q SOP are the active power and reactive power dispatched by SOP respectively; β1, β2, β3 and c1, c2, c3 are price constants respectively; P chr , Q chr They are respectively the active power and reactive power output values ​​of the energy storage equipment; In the formula, is the system voltage variance; ξ1, ξ2, ξ3 are voltage offset penalty parameters.

5. The intelligent soft switching method for operation and dispatching of active distribution network of highway according to claim 1, characterized in that: S3.

2. In the construction of the second-order optimization algorithm and algorithm constraints, the NSGA-II algorithm is used as the upper-level optimization algorithm to solve the switching strategy for SOP settings in each time period; when faced with continuous variables and linear constraints such as power scheduling, the PSO algorithm is used to solve the scheduling power value.

Citation Information

Patent Citations

  • Power distribution network robust recovery decision-making method considering uncertainty of distributed power supply

    CN111478358A

  • Active power distribution network and multi-microgrid joint robust optimization method and system

    CN114726008A

  • Active power distribution network expansion planning method and system considering intelligent soft switch access

    CN114996908A

  • Power distribution network active-reactive joint optimization method for coordinating network side resources

    CN116388302A