Voltage hierarchical optimization method and system under cloud edge interaction control

Abstract

The invention discloses a voltage hierarchical optimization method and system under cloud-side interaction control, and the method comprises the following steps: building a medium-voltage distribution network cloud-side interaction voltage control frame based on the regulation and control of a distributed power distribution network on reactive voltage; correspondingly establishing a distributed layered reactive voltage control model based on the control framework; using multivariate normal distribution to randomly generate a scene data set of reactive voltage control, and training a control model; and solving the control model by using a clustering algorithm to obtain a distributed power distribution network output scheme of the reactive voltage node to be optimized. According to the method, the in-situ layer is matched with the execution layer and the coordination layer, a data processing part is deployed in a cloud system, an integrated intelligent agent is provided, a basis is provided for supporting the flexibility of cloud edge coordination regulation and control, reactive power compensation equipment is added or reduced according to the obtained output scheme, operation is simple, and implementation is easy.

Description

Technical Field

[0001] The present invention belongs to the field of electric power technology, relates to reactive voltage regulation technology, and is a voltage stratification optimization method under cloud-edge interactive control. Background Art

[0002] The access of distributed power sources in distribution networks is increasingly widely used, and traditional distribution networks have gradually transformed into active distribution networks. However, the widespread application of distributed distribution systems has brought great challenges to the voltage and reactive power control of traditional distribution networks. For example, photovoltaic generators at the end of feeders often violate voltage limits. For the coordinated control of voltage and reactive power, the use of distributed power sources and traditional voltage regulation equipment is particularly important. Considering the reactive characteristics of the inverter, the inverter can be controlled to adapt to the rapid fluctuations of the new distributed distribution network on a fast and short time scale of seconds, thereby greatly reducing the reactive compensation equipment required for the distribution network. In active distribution networks, the effective coordination of distributed power sources in voltage control is a crucial issue. At the same time, voltage control usually adopts a centralized control method, but it often encounters major challenges in large-scale practical applications. In order to achieve more effective distributed control, a reasonable voltage optimization method is of great significance for the safe and reliable operation of the power grid.

[0003] After searching the prior art, it was found that the Chinese patent application CN117913843A, "A reactive voltage control method based on an improved firefly swarm optimization algorithm", described a control variable sequence that includes multiple parameters, including the terminal voltage of the active-voltage type thermal power unit, the transformer ratio, the reactive output of the SVC, etc., but the patent did not indicate the coordinated relationship between the voltages of each node. The Chinese patent application CN117878954A, "A voltage optimization method and device based on distributed photovoltaics", characterized a voltage optimization model constructed by a reactive-voltage droop function, but did not further propose a solution algorithm for the voltage optimization model. Summary of the invention

[0004] The technical problem to be solved by the present invention is that the existing methods for evaluating the resilience of distribution networks are insufficient in comprehensiveness and operability.

[0005] The technical solution of the present invention is: a voltage stratification optimization method under cloud-edge interactive control, comprising the following steps:

[0006] S1: Based on the regulation of reactive voltage by distributed distribution network, a cloud-edge interactive voltage control framework for medium-voltage distribution network is constructed. The control framework for cloud-edge interactive voltage in medium-voltage distribution network is an integrated intelligent entity, which adopts cloud-edge interactive control. The data processing part is deployed in the cloud system. The control framework is divided into three levels, namely coordination layer, execution layer and local layer. The coordination layer corresponds to the operation of the transmission system, the execution layer corresponds to the operation of the distribution system, and the local layer corresponds to the distributed output equipment. The coordination layer is used for the deployment of reactive voltage compensation, and the execution layer is used to issue operation instructions to the local layer.

[0007] S2: Based on the cloud-edge interactive voltage control framework of the medium-voltage distribution network, a distributed hierarchical reactive voltage control model is established corresponding to the coordination layer, execution layer and local layer, and the objective functions and constraints of the coordination layer and local layer are constructed;

[0008] S3: Use multivariate normal distribution to randomly generate scenario data sets of reactive voltage control, which are used to train the distributed hierarchical reactive voltage control model and learn the impact of reactive voltage on the distribution network and the response of the distribution network under various scenarios;

[0009] S4: Determine the reactive voltage node to be optimized, and use the clustering algorithm to iteratively solve the trained distributed hierarchical reactive voltage control model based on the objective function and constraints. The clustering results are combined with the corresponding output plan to obtain the distributed distribution network optimized output plan of the reactive voltage node to be optimized. The optimized output of the distribution network is realized through the execution layer, and reactive power compensation equipment is increased or decreased.

[0010] Further, in step S2, the coordination layer and the local layer of the distributed reactive power voltage control model are respectively:

[0011] The coordination layer and the local layer cooperate with each other, and each parameter meets the operation constraints of the distribution network, which include the balance of active power, reactive power and voltage amplitude;

[0012] Considering the coordinated dispatch of the transmission system, the objective function of the reactive power and voltage optimization model at the coordination layer is expressed as:

[0013]

[0014] In the formula, i represents the reactive voltage node number; G represents the set of all reactive voltage nodes; t represents the time period of dispatch output; K represents the set of all time periods; σ i,t,loss It is represented as the network loss of node i at time t; E is represented as the set of distribution network branches pq; pq is the branch from node p to node q; L is the current flowing through branch pq, and R is the total resistance of branch pq;

[0015] In order to realize interactive control between the coordination layer and the local layer, and considering the difference between the grid loss dimension and the voltage dimension, the objective function of the coordination layer is normalized, and the objective function of the reactive power voltage optimization model of the local layer is expressed as:

[0016]

[0017] In the formula, and is the node network loss and node voltage deviation after normalization; β is the node voltage deviation coefficient, and its value is between 0 and 1.

[0018] Furthermore, in step S3, the specific steps of randomly generating scenario data for reactive power and voltage control using multivariate normal distribution are as follows:

[0019] S3-1: Establishing random vectors of reactive power and time In the formula, the output time point t 1 =99;

[0020] S3-2: Assume that the random vector Q follows a multivariate normal distribution, Q~N(E 0 ,Σ), expect E 0 is a 99-dimensional zero vector, and the matrix Σ is represented as:

[0021]

[0022] In the formula, the coefficient λ in the matrix is ​​the correlation coefficient between the random vector Q and the node network loss, and its value is greater than 1;

[0023] S3-3: Combine the multivariate normal distribution matrix and random vectors to generate random scenarios of reactive output using the mvnrnd function in MATLAB;

[0024] S3-4: Obtain random scenario data of reactive power and voltage control considering time correlation.

[0025] Furthermore, in step S4, a clustering algorithm is used to cluster the reactive voltage node data set to be optimized, and the calculation formula is:

[0026]

[0027] In the formula, x is the cluster center node before calculation; y is the total number of center nodes; S m For all clusters O m One of them; M mn and N mn Represents the nodes and central nodes of the mth cluster; u Om Represented as cluster O mThe total number of central nodes in the group; the cluster center is continuously replaced until all cluster nodes are replaced or the value of the objective function has been minimized, the clustering result is determined, and the reactive voltage configuration and voltage regulation scheme of each cluster are integrated to form the overall output plan of the distribution network.

[0028] Furthermore, each level of the cloud-edge interactive voltage control framework of the medium-voltage distribution network is configured with a relatively independent optimization model.

[0029] The present invention also proposes a voltage stratification optimization system under cloud-edge interactive control, which configures the cloud, edge and edge devices of the cloud-edge collaborative system according to the coordination layer, execution layer and local layer of the above-mentioned control framework, and implements the above-mentioned voltage stratification optimization method under cloud-edge interactive control, outputs a distributed distribution network optimization output plan for the reactive voltage nodes to be optimized in the distribution network, and realizes reactive voltage regulation of the distribution network.

[0030] The method proposed in the present invention constructs an integrated intelligent entity with three levels, taking into account the influence of multiple time scales, and can give full play to the characteristics of distributed power sources in the active distribution network, ensuring that the power system has strong and effective reactive voltage regulation capabilities in emergency situations. It is suitable for large-scale active distribution networks with extensive access to distributed power sources, and is of great significance for the reliable reserve of reactive voltage in the system. Operators can use this control framework to provide the system with reasonable reactive voltage support.

[0031] Compared with the prior art, the present invention has the following beneficial effects:

[0032] 1. Compared with the prior art, the present invention fully considers the advantages of the distribution network containing distributed power sources that can flexibly regulate reactive voltage, coordinates the local layer with the execution layer and the coordination layer, deploys the data output part in the cloud system, and proposes an integrated intelligent body, which provides a basis for supporting the flexibility of cloud-edge collaborative regulation.

[0033] 2. The present invention adopts a voltage stratification optimization method under cloud-edge interactive control, takes into account the coordinated scheduling of the transmission system, avoids the decentralization of data transmission between levels, improves the economy of reactive voltage optimization and reduces the time delay between data transmission, which plays an important role in solving the real-time and rapid response to reactive voltage shortage. BRIEF DESCRIPTION OF THE DRAWINGS

[0034] Figure 1 It is a schematic diagram of the process of the present invention. DETAILED DESCRIPTION

[0035] The following will clearly and completely describe the technical solutions in the embodiments of the present invention in conjunction with the accompanying drawings. Obviously, the described embodiments are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0036] like Figure 1 As shown, a voltage stratification optimization method under cloud-edge interactive control includes the following steps.

[0037] S1: Considering the advantage of being able to flexibly regulate reactive voltage in distributed distribution networks, a cloud-edge interactive voltage control framework for medium-voltage distribution networks is proposed; the control framework for cloud-edge interactive voltage in the medium-voltage distribution network is an integrated intelligent entity, which adopts cloud-edge interactive control, and the data processing part is deployed in the cloud system. The control framework is divided into three levels, namely, coordination layer, execution layer and local layer. The coordination layer corresponds to the operation of the transmission system, the execution layer corresponds to the operation of the distribution system, and the local layer corresponds to the distributed output equipment. The coordination layer is used for the deployment of reactive voltage compensation, and the execution layer is used to issue operation instructions to the local layer.

[0038] S2: Based on the cloud-edge interactive voltage control framework of the medium-voltage distribution network, a distributed hierarchical reactive voltage control model is established corresponding to the coordination layer, execution layer and local layer, and the objective functions and constraints of the coordination layer and local layer are constructed.

[0039] S3: Use multivariate normal distribution to randomly generate scenario data sets for reactive voltage control to train a distributed hierarchical reactive voltage control model. In order to improve the accuracy of the control model, the present invention uses multivariate normal distribution to randomly generate scenario data sets for reactive voltage control as training sets, thereby training a distributed hierarchical reactive voltage model that can understand and respond to complex scenarios; then, in actual applications, by inputting different scenario data, the corresponding optimization solution is output under the objective function.

[0040] S4: Determine the reactive voltage node to be optimized, and use the clustering algorithm to iteratively solve the trained distributed hierarchical reactive voltage control model based on the objective function and constraints. The distributed distribution network optimized output scheme of the reactive voltage node to be optimized is obtained by combining the clustering results with the corresponding output scheme. The optimized output of the distribution network is realized through the execution layer, and the reactive power compensation equipment is increased or decreased. When the distributed hierarchical reactive voltage model is solved, the clustering algorithm can help explain the output of the neural network. The distributed hierarchical reactive voltage model may solve and output multiple schemes under the objective function, containing a large amount of information, some of which may be redundant or less important. The present invention optimizes and screens these outputs through the clustering algorithm. The clustering algorithm clusters the reactive voltage node data set to be optimized. After determining the clustering results, the reactive voltage configuration and voltage regulation scheme of each cluster are integrated to form an overall output scheme of the distribution network, and the distribution network voltage is optimized through the execution layer.

[0041] In step S2 of this embodiment, the coordination layer and the local layer of the distributed reactive power voltage control model are respectively:

[0042] The coordination layer and the local layer cooperate with each other, and each parameter meets the operating constraints of the distribution network, including the balance of active power, reactive power and voltage amplitude.

[0043] Considering the coordinated dispatch of the transmission system, the objective function of the reactive power and voltage optimization model at the coordination layer is expressed as:

[0044]

[0045] In the formula, i represents the reactive voltage node number in the distribution network; G represents the set of all nodes; t represents the time period of dispatching output; K represents the set of all time periods; σ i,t,loss It is represented as the network loss of node i at time t; E is represented as the set of branches pq; pq is the branch from node p to node q; L is the current flowing through branch pq, and R is the total resistance of branch pq.

[0046] In order to realize interactive control between the coordination layer and the local layer, and considering the difference between the grid loss dimension and the voltage dimension, the objective function of the coordination layer is normalized, including normalization and regularization. The objective function of the reactive voltage optimization model of the local layer is expressed as:

[0047]

[0048] In the formula, and is the node network loss and node voltage deviation after normalization; β is the node voltage deviation coefficient, and its value is between 0 and 1.

[0049] In this embodiment, the specific steps of randomly generating scenario data of reactive power and voltage control using multivariate normal distribution are as follows:

[0050] S3-1: Establishing random vectors of reactive power and time In the formula, the output time point t 1 =99;

[0051] S3-2: Assume that the random vector Q follows a multivariate normal distribution, Q~N(E 0 ,Σ), expect E 0 is a 99-dimensional zero vector, and the matrix Σ is represented as:

[0052]

[0053] Wherein, the coefficient λ in the matrix is ​​the correlation coefficient between the random vector Q and the node network loss, and its value is greater than 1.

[0054] S3-3: Combine the multivariate normal distribution matrix and random vectors to generate random scenarios of reactive output using the mvnrnd function in MATLAB;

[0055] S3-4: Obtain random scenario data of reactive power and voltage control considering time correlation.

[0056] In step S4 of this embodiment, a clustering algorithm is used to assist the distributed hierarchical reactive voltage control model in solving. The advantage of using a clustering algorithm is that only the center of each cluster node needs to be replaced to complete an iterative solution process. The calculation formula is:

[0057]

[0058] In the formula, x is the cluster center node before calculation; y is the total number of center nodes; S m For all clusters O m One of them; M mn and N mn Represents the nodes and central nodes of the mth cluster; u Om Represented as cluster O m The total number of central nodes in the group; the cluster center is continuously replaced until all cluster nodes are replaced or the value of the objective function has been minimized, the clustering result is determined, the reactive voltage configuration and voltage regulation scheme of each cluster are integrated to form the overall output plan of the distribution network, which is executed through the execution layer to increase or decrease reactive power compensation equipment.

[0059] Based on the above description, the present invention also proposes a voltage hierarchical optimization system under cloud-edge interactive control, which configures the cloud, edge and edge devices of the cloud-edge collaborative system according to the coordination layer, execution layer and local layer of the above control framework, and implements the voltage hierarchical optimization method under the above cloud-edge interactive control, outputs a distributed distribution network optimization output plan for the reactive voltage node to be optimized in the distribution network, and realizes reactive voltage regulation of the distribution network. The present invention adopts cloud-edge interactive control to avoid decentralized transmission. For the coordinated scheduling of the transmission system, it can improve the economy of reactive voltage optimization and reduce the time delay between data transmission.

Claims

1. A voltage stratification optimization method under cloud-edge interactive control, characterized in that: The following steps are involved: S1: Based on the regulation of reactive voltage by distributed distribution network, a cloud-edge interactive voltage control framework for medium-voltage distribution network is constructed. The control framework for cloud-edge interactive voltage in medium-voltage distribution network is an integrated intelligent entity, which adopts cloud-edge interactive control. The data processing part is deployed in the cloud system. The control framework is divided into three levels, namely coordination layer, execution layer and local layer. The coordination layer corresponds to the operation of the transmission system, the execution layer corresponds to the operation of the distribution system, and the local layer corresponds to the distributed output equipment. The coordination layer is used for the deployment of reactive voltage compensation, and the execution layer is used to issue operation instructions to the local layer. S2: Based on the cloud-edge interactive voltage control framework of the medium-voltage distribution network, a distributed hierarchical reactive voltage control model is established corresponding to the coordination layer, execution layer and local layer, and the objective functions and constraints of the coordination layer and local layer are constructed; S3: Use multivariate normal distribution to randomly generate scenario data sets of reactive voltage control, which are used to train the distributed hierarchical reactive voltage control model and learn the impact of reactive voltage on the distribution network and the response of the distribution network under various scenarios; S4: Determine the reactive voltage node to be optimized, and use the clustering algorithm to iteratively solve the trained distributed hierarchical reactive voltage control model based on the objective function and constraints. The clustering results are combined with the corresponding output plan to obtain the distributed distribution network optimized output plan of the reactive voltage node to be optimized. The optimized output of the distribution network is realized through the execution layer, and reactive power compensation equipment is increased or decreased.

2. The voltage stratification optimization method under cloud-edge interactive control according to claim 1 is characterized in that: In step S2, the coordination layer and local layer of the distributed reactive power voltage control model are: The coordination layer and the local layer cooperate with each other, and each parameter meets the operation constraints of the distribution network, which include the balance of active power, reactive power and voltage amplitude; Considering the coordinated dispatch of the transmission system, the objective function of the reactive power and voltage optimization model at the coordination layer is expressed as: In the formula, i represents the reactive voltage node number; G represents the set of all reactive voltage nodes; t represents the time period for dispatching output; K represents the set of all time periods; σ i,t,loss It is represented as the network loss of node i at time t; E is represented as the set of distribution network branches pq; pq is the branch from node p to node q; L is the current flowing through branch pq, and R is the total resistance of branch pq; In order to realize interactive control between the coordination layer and the local layer, and considering the difference between the grid loss dimension and the voltage dimension, the objective function of the coordination layer is normalized, and the objective function of the reactive power voltage optimization model of the local layer is expressed as: In the formula, and is the node network loss and node voltage deviation after normalization; β is the node voltage deviation coefficient, and its value is between 0 and 1.

3. The voltage stratification optimization method under cloud-edge interactive control according to claim 1 is characterized in that: In step S3, the specific steps of randomly generating scenario data for reactive power and voltage control using multivariate normal distribution are as follows: S3-1: Establishing random vectors of reactive power and time In the formula, the output time point t1 = 99; S3-2: Assume that the random vector Q obeys the multivariate normal distribution, Q~N(E0,Σ), and expect E0 to be a 99-dimensional zero vector. The matrix Σ is expressed as: In the formula, the coefficient λ in the matrix is ​​the correlation coefficient between the random vector Q and the node network loss, and its value is greater than 1; S3-3: Combine the multivariate normal distribution matrix and random vectors to generate random scenarios of reactive output using the mvnrnd function in MATLAB; S3-4: Obtain random scenario data of reactive power and voltage control considering time correlation.

4. The voltage stratification optimization method under cloud-edge interactive control according to claim 1 is characterized in that: In step S4, a clustering algorithm is used to cluster the reactive voltage node data set to be optimized, and the calculation formula is: In the formula, x is the cluster center node before calculation; y is the total number of center nodes; S m For all clusters O m One of them; M mn and N mn Represents the nodes and central nodes of the mth cluster; u Om Represented as cluster O m The total number of central nodes in the group; the cluster center is continuously replaced until all cluster nodes are replaced or the value of the objective function has been minimized, the clustering result is determined, and the reactive voltage configuration and voltage regulation scheme of each cluster are integrated to form the overall output plan of the distribution network.

5. The voltage stratification optimization method under cloud-edge interactive control according to claim 1 is characterized in that: Each level of the cloud-edge interactive voltage control framework of the medium-voltage distribution network is configured with a relatively independent optimization model.

6. A voltage stratification optimization system under cloud-edge interactive control, characterized in that: The cloud, edge and edge devices of the cloud-edge collaborative system are configured according to the coordination layer, execution layer and local layer of the control framework described in any one of claims 1-5, and the voltage layered optimization method under the cloud-edge interactive control described in any one of claims 1-5 is implemented, and the distributed distribution network optimization output plan is output for the reactive voltage nodes to be optimized in the distribution network, so as to realize reactive voltage regulation of the distribution network.

Images