Active power distribution network voltage weak point identification method considering photovoltaic output

Abstract

The invention discloses an active power distribution network voltage weak point identification method considering photovoltaic output, and solves the problems that node voltage frequently exceeds a limit and the weakness of a voltage node is enhanced under the background that large-scale photovoltaic access to an active power distribution network. According to the method, the photovoltaic output condition is predicted by establishing a neural network model, and then the voltage deviation ratio and the voltage sensitivity parameter of each node are obtained through load flow calculation. And on the basis, network structure characteristic parameters are calculated in combination with power distribution network topology, an active power distribution network comprehensive voltage weak point index system is constructed, the voltage weak degree of each node is quantified, and accurate identification of active power distribution network voltage weak points is realized. According to the method, the randomness of photovoltaic output and the comprehensiveness of various indexes are considered, the influence of different illumination scenes on the voltage weak point of the active power distribution network can be obtained, the adaptability of the identification method to an actual operation scene is improved, and decision support can be provided for safe and stable operation and scientific planning design of the active power distribution network.

Description

Technical Field

[0001] This invention relates to a method for identifying voltage weak points in active power distribution networks that takes into account photovoltaic output, and belongs to the field of power quality monitoring technology for active power distribution networks. Background Technology

[0002] In recent years, with the rapid development of new energy technologies, a large number of distributed photovoltaic (PV) and other renewable energy sources have been connected to active power distribution networks. However, due to the significant fluctuations and randomness in PV output caused by weather, seasons, and environmental factors, voltage stability issues in distribution networks have become increasingly prominent. This is especially true in scenarios with high PV penetration, where frequent voltage exceedances at local nodes occur, seriously threatening the safe and stable operation of the power system. Therefore, accurately identifying and assessing voltage-weak nodes in active power distribution networks has become one of the key issues for the stable operation and scientific planning of new power systems.

[0003] Traditional methods for identifying voltage weaknesses in active distribution networks are typically based on deterministic load and power source models, employing sensitivity analysis and voltage margin calculations. However, these methods fail to consider the randomness and uncertainty of photovoltaic (PV) output, making it difficult to accurately identify voltage weaknesses in active distribution networks. Furthermore, existing weakness identification indicators often rely on the electrical quantities of nodes, failing to fully integrate the topological characteristics of network nodes, resulting in somewhat one-sided identification results. Therefore, there is an urgent need to propose a method for identifying voltage weaknesses in active distribution networks that can effectively characterize the uncertainty of PV output and comprehensively consider the electrical characteristics of nodes and network structure features. Summary of the Invention

[0004] The technical problem to be solved by this invention is to provide a method for identifying voltage weak points in active distribution networks that takes into account photovoltaic output. By introducing a clustering and prediction mechanism for photovoltaic output and combining network topology characteristics and node electrical characteristics, a comprehensive weak point evaluation index is constructed to achieve efficient and accurate identification of voltage weak points in active distribution networks.

[0005] To solve the above-mentioned technical problems, the present invention adopts the following technical solution:

[0006] A method for identifying voltage weak points in an active power distribution network that considers photovoltaic output includes the following steps:

[0007] Step 1: For the active distribution network with weak voltage points to be identified, obtain historical weather data of the area where the active distribution network is located and preprocess it. Divide the preprocessed historical weather data into three categories, corresponding to three lighting scenarios: sunny, cloudy and rainy days.

[0008] Step 2: Obtain the photovoltaic output of the active power distribution network corresponding to each historical weather under each illumination scenario, and establish a long short-term memory neural network photovoltaic output prediction model; for each illumination scenario, take the preprocessed historical weather data as input and the corresponding active power distribution network photovoltaic output as output to train the photovoltaic output prediction model corresponding to each illumination scenario.

[0009] Step 3: Perform power flow calculations on the active distribution network using the Newton-Layer method to obtain the electrical characteristic indicators of each node in the active distribution network, including voltage deviation rate and voltage sensitivity. During the power flow calculation, the injected active power of the photovoltaic output nodes in the active distribution network is predicted using the photovoltaic output prediction model trained in Step 2, while the injected active power of other nodes is known.

[0010] Step 4: Combine the active distribution network topology information to calculate the structural characteristic indicators of each node, including node degree, betweenness centrality, and average path length.

[0011] Step 5: Based on the node indicators obtained in Step 3 and Step 4, normalize each indicator. Calculate the comprehensive voltage weakness index value of each node based on the normalized indicators. Nodes with comprehensive voltage weakness index values ​​greater than a preset threshold are identified as voltage weakness nodes, thereby realizing the identification of voltage weakness points in the active distribution network.

[0012] Compared with the prior art, the present invention, employing the above technical solution, has the following technical effects:

[0013] 1. This invention divides photovoltaic power generation into three typical illumination scenarios through cluster analysis and combines them with a neural network model to predict photovoltaic power generation data. It fully considers the randomness and volatility of photovoltaic output and improves the accuracy and dynamic adaptability of identifying weak nodes in the active power distribution network.

[0014] 2. This invention establishes a voltage node vulnerability index system that integrates two key influencing factors: electrical characteristics and network topology features. It comprehensively evaluates the vulnerability of nodes in the power grid, improves the accuracy and practicality of identifying vulnerable nodes in active distribution networks, and has higher identification accuracy and comprehensiveness compared with traditional methods.

[0015] 3. This invention uses normalization and entropy weighting to uniformly evaluate various indicators, overcoming the evaluation bias caused by inconsistent indicator dimensions, making the ranking of voltage weak points more reasonable and possessing good stability and robustness.

[0016] 4. This invention is applicable to active distribution network scenarios with a high proportion of distributed photovoltaic access, and can provide theoretical support and decision-making basis for voltage regulation, network structure optimization and distribution network planning, and has broad engineering application prospects. Attached Figure Description

[0017] Figure 1 This is a flowchart of a method for identifying voltage weak points in an active power distribution network that takes into account photovoltaic output, according to the present invention.

[0018] Figure 2 This is a simple power distribution network topology diagram;

[0019] Figure 3 This is a diagram of the long short-term memory neural network structure;

[0020] Figure 4 This is a diagram of the training data format;

[0021] Figure 5 This is a ranking diagram of the voltage weakness of each node. Detailed Implementation

[0022] Embodiments of the present invention are described in detail below, examples of which are illustrated in the accompanying drawings. The embodiments described below with reference to the accompanying drawings are exemplary and are only used to explain the present invention, and should not be construed as limiting the present invention.

[0023] like Figure 1 As shown, this invention proposes a method for identifying voltage weaknesses in active power distribution networks that considers photovoltaic output. This includes calculating the electrical characteristic indicators of nodes considering photovoltaic output and calculating a comprehensive node weakness index that integrates electrical characteristics and network topology features. The specific process is as follows:

[0024] Step 1: Perform cluster analysis on historical weather data (sunlight intensity, temperature, humidity) to divide photovoltaic power output into three typical sunlight scenarios: sunny days, cloudy days, and rainy days.

[0025] Figure 2 The diagram shows a simple 14-node distribution network topology, with photovoltaic (PV) systems connected at nodes 3, 5, 7, and 9. First, 1000 days of historical weather data for a specific location are selected, including data on sunlight, temperature, and humidity. The historical weather data represents the average of the current day's data. A weather vector X is constructed as follows:

[0026] X = [x1, x2, ..., x i ,...,x n ]

[0027] x i =[I i ,T i H i ]

[0028] In the formula, i represents the weather data for the i-th day, I i T represents the average daily light intensity. i H represents the average temperature at that time. iThis represents the average humidity of the day, n = 1000.

[0029] After normalizing the weather vector X, the K-means clustering algorithm was used to divide the selected 1000 days of historical weather data into three categories, corresponding to three typical lighting scenarios. First, k (3 in this invention) initial cluster centers C were randomly selected from the weather vector X. j (1≤j≤k), calculate the relationship between the remaining data objects and the cluster center C respectively. j The Euclidean distance is used to assign the data objects with the smallest Euclidean distance to the cluster center C. j Within the corresponding clusters, the average value of the data objects in each cluster is calculated as the new cluster center, and the next iteration is performed until the cluster centers no longer change or the maximum number of iterations is reached.

[0030] The formula for calculating the Euclidean distance between data objects and cluster centers in space is:

[0031]

[0032] In the formula, x i For data objects; C j Let be the j-th cluster center; p is the dimension of the data object, which in this example is three-dimensional data of light, temperature, and humidity.

[0033] Step 2: Establish a neural network model with long short-term memory neural network as the core, train it on historical photovoltaic power generation data for each typical lighting scenario, and predict photovoltaic power generation data for each lighting scenario.

[0034] After obtaining three typical lighting scenes, establish separate systems for each scene. Figure 3 The neural network structure shown is used for model training to improve the accuracy of photovoltaic power output prediction and enhance adaptability to different scenarios. The core of the entire network structure consists of two LSTM hidden layers and three fully connected layers. A normalization layer is added in the middle to accelerate training convergence and enhance model stability, and a dropout layer (dropout rate of 0.5) is added to prevent overfitting.

[0035] The neural network is trained using the day's weather data as input and outputs the predicted photovoltaic power output for that day, such as... Figure 4 As shown in the figure. The photovoltaic output data sampling interval is 15 minutes, with a total of 96 data points. Therefore, the historical data of 1000 days is reconstructed into a 1000*99 training matrix, where the first three columns are the input and the last 96 columns are the output for neural network training. To accelerate the network training speed, a mini-batch training method is adopted, with a mini-batch size of 25 and a training iteration count of 50.

[0036] Step 3: Based on the predicted photovoltaic power generation data, the Newton-Layer method is used to perform power flow calculations for each typical illumination scenario to obtain the voltage deviation rate and voltage sensitivity parameters of each node.

[0037] After the photovoltaic output of nodes 3, 5, 7, and 9 is predicted using the trained neural network, the power flow calculation of the distribution network is performed using the Newton-Layer method.

[0038] The power flow equations for each node are defined as follows:

[0039]

[0040] In the formula, P m Q is the injected active power at node m. m U is the reactive power injected into node m. m U is the voltage magnitude at node m. n Let θ be the voltage magnitude at node n. mn G represents the phase angle difference between node m and node n. mn Let B be the real admittance between node m and node n. mn Let be the imaginary admittance between node m and node n, where n represents the node connected to node m.

[0041] The above system of equations is established for all voltage nodes, and solved iteratively using the Jacobian matrix:

[0042]

[0043] Repeat the above iterative solution until the error meets the convergence condition to obtain the voltage parameters of each node in the distribution network system. Based on the node voltages, calculate the voltage deviation rate of each node.

[0044]

[0045] In the formula, V m U represents the voltage deviation rate at node m. N This indicates the rated voltage.

[0046] To further reflect the voltage response capability of photovoltaic access nodes to power changes, the following voltage sensitivity index is determined:

[0047]

[0048] The greater the voltage sensitivity, the greater the weakness of the node.

[0049] Step 4: Based on this, and in conjunction with the power distribution network topology information, calculate the structural characteristic indicators of each node, including node degree, average path length, and betweenness centrality parameter.

[0050] in accordance with Figure 2The diagram shows a simple power distribution network topology. The node degree, betweenness centrality, and average path length of each node are calculated as follows: Node degree is represented by the number of directly connected neighboring nodes. A smaller node degree indicates a smaller impact of changes in the node's operating state on neighboring nodes. Therefore, a smaller node degree indicates a weaker node and weaker voltage support. The calculation method is as follows:

[0051]

[0052] In the formula, D m Let f be the degree of node m, and let f represent any other node in the system besides node m. If node m and f are connected, then a mf =1, otherwise it is 0.

[0053] Betweenness centrality indicates the frequency at which a node appears on all shortest electrical impedance paths, reflecting its ability to control power flow. A higher betweenness centrality indicates a greater impact of the node on the entire network, a higher risk of problems, and therefore a greater voltage vulnerability. The calculation method is as follows:

[0054]

[0055] In the formula, B m For the betweenness centrality of node m, g st This represents the number of minimum impedance paths between nodes s and t, typically 1. When there are h equal minimum paths, h is taken as h. σ st (m) represents the number of nodes m that these paths pass through.

[0056] The average path length represents the average of the shortest electrical impedance path lengths from this node to all other nodes. A longer average path length indicates a weaker voltage support capability and lag in regulation at that node, thus resulting in a more vulnerable voltage condition. The calculation method is as follows:

[0057]

[0058] In the formula, L m Let be the average path length of node m, N be the number of nodes in the system, and l(m,f) represent the shortest path length from node m to node f.

[0059] Step 5: Construct a comprehensive voltage weakness index that includes node electrical characteristic indicators and network structure indicators, and sort the weakness indicators of all nodes in the network to achieve accurate identification of voltage weaknesses in the active distribution network.

[0060] After obtaining the electrical characteristic indicators of each node, in order to more comprehensively consider the impact of the distribution network topology characteristics on voltage vulnerability, this invention proposes a comprehensive node vulnerability index calculation method that integrates node electrical characteristics and network topology characteristics, as follows:

[0061] After calculating the electrical parameters of all nodes—voltage deviation rate and voltage sensitivity—and the structural characteristic parameters—node degree, betweenness centrality, and average path length—assuming the system has N nodes, the parameter matrix is ​​constructed as follows:

[0062]

[0063] Normalize the matrix Z column-wise for each index. Since not all indices are positively correlated with voltage weakness (e.g., node degree is negatively correlated with voltage weakness), the normalized node degree index is processed by setting D' = 1 - D. Then all indices are positively correlated with voltage weakness. The comprehensive voltage weakness index for each node is calculated as follows:

[0064] W=αV+βS+γD'+δB+εL

[0065] In the formula, α, β, γ, δ, and ε are the weighting coefficients of each index, calculated using the entropy weighting method. Taking α as an example, the proportion of each node in the voltage deviation rate index is first calculated:

[0066]

[0067] Next, the information entropy of the voltage deviation rate index is calculated as follows:

[0068]

[0069] The entropy redundancy calculated from the above formula is:

[0070] d V =1-e V

[0071] The weighting coefficient for the voltage deviation rate index is:

[0072]

[0073] In the formula, d V d S d D d B and d L These are the entropy redundancy of voltage deviation rate, voltage sensitivity, nodality, betweenness centrality, and average path length, respectively.

[0074] For the comprehensive voltage vulnerability index of each node, the larger the index, the greater the voltage vulnerability of the node. The nodes are sorted in descending order to obtain... Figure 5 The diagram shows the ranking of voltage weaknesses at each node. A threshold value is set, and nodes in the power distribution system whose overall voltage weakness index exceeds the threshold are identified as voltage weaknesses. The threshold value is selected based on engineering experience and is generally set to 0.6.

[0075] Based on the same inventive concept, this application provides a computer device, including a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, it implements the steps of the aforementioned method for identifying weak points in the voltage of an active power distribution network that takes into account photovoltaic output.

[0076] Based on the same inventive concept, embodiments of this application provide a computer-readable storage medium storing a computer program that, when executed by a processor, implements the steps of the aforementioned method for identifying weak points in the voltage of an active power distribution network that considers photovoltaic output.

[0077] Those skilled in the art will understand that embodiments of the present invention can be provided as methods, systems, or computer program products. Therefore, the present invention can take the form of a completely hardware embodiment, a completely software embodiment, or an embodiment combining software and hardware aspects. Furthermore, the present invention can take the form of a computer program product embodied on one or more computer-usable storage media (including, but not limited to, disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0078] This invention is described with reference to flowchart illustrations and / or block diagrams of methods, apparatus (systems), and computer program products according to embodiments of the invention. It will be understood that each block of the flowchart illustrations and / or block diagrams, and combinations of blocks in the flowchart illustrations and / or block diagrams, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, special-purpose computer, embedded processor, or other programmable data processing apparatus to produce a machine, such that the instructions, which execute via the processor of the computer or other programmable data processing apparatus, generate instructions for implementing the flowchart illustrations and / or block diagrams. Figure 1 One or more processes and / or boxes Figure 1 A device that provides the functions specified in one or more boxes.

[0079] These computer program instructions may also be stored in a computer-readable storage medium that can direct a computer or other programmable data processing device to function in a particular manner, such that the instructions stored in the computer-readable storage medium produce an article of manufacture including instruction means, which are implemented in a process Figure 1One or more processes and / or boxes Figure 1 The function specified in one or more boxes.

[0080] These computer program instructions may also be loaded onto a computer or other programmable data processing equipment to cause a series of operational steps to be performed on the computer or other programmable equipment to produce a computer-implemented process, thereby providing instructions that execute on the computer or other programmable equipment for implementing the process. Figure 1 One or more processes and / or boxes Figure 1 The steps of the function specified in one or more boxes.

[0081] The above embodiments are merely illustrative of the technical concept of the present invention and should not be construed as limiting the scope of protection of the present invention. Any modifications made to the technical solutions based on the technical concept proposed in this invention shall fall within the scope of protection of this invention.

Claims

1. A method for identifying voltage weak points in an active power distribution network considering photovoltaic output, characterized in that, Includes the following steps: Step 1: For the active distribution network with weak voltage points to be identified, obtain historical weather data of the area where the active distribution network is located and preprocess it. Divide the preprocessed historical weather data into three categories, corresponding to three lighting scenarios: sunny, cloudy and rainy days. Step 2: Obtain the photovoltaic output of the active power distribution network corresponding to each historical weather under each illumination scenario, and establish a long short-term memory neural network photovoltaic output prediction model; for each illumination scenario, take the preprocessed historical weather data as input and the corresponding active power distribution network photovoltaic output as output to train the photovoltaic output prediction model corresponding to each illumination scenario. Step 3: Perform power flow calculations on the active distribution network using the Newton-Layer method to obtain the electrical characteristic indicators of each node in the active distribution network, including voltage deviation rate and voltage sensitivity. During the power flow calculation, the injected active power of the photovoltaic output nodes in the active distribution network is predicted using the photovoltaic output prediction model trained in Step 2, while the injected active power of other nodes is known. Step 4: Combine the active distribution network topology information to calculate the structural characteristic indicators of each node, including node degree, betweenness centrality, and average path length. Step 5: Based on the node indicators obtained in Step 3 and Step 4, normalize each indicator. Calculate the comprehensive voltage weakness index value of each node based on the normalized indicators. Nodes with comprehensive voltage weakness index values ​​greater than a preset threshold are identified as voltage weakness nodes, thereby realizing the identification of voltage weakness points in the active distribution network.

2. The method for identifying weak points in the voltage of an active power distribution network considering photovoltaic output as described in claim 1, characterized in that, In step 1, the historical weather data includes the average light intensity, average temperature, and average humidity data for each day; the average light intensity, average temperature, and average humidity data for each day are preprocessed by normalization. The K-means clustering algorithm was used to divide the normalized preprocessed historical weather data into corresponding lighting scenarios for sunny, cloudy, or rainy days.

3. The method for identifying weak points in the voltage of an active power distribution network considering photovoltaic output according to claim 1, characterized in that, In step 2, the long short-term memory neural network photovoltaic power output prediction model includes an input layer, a first LSTM hidden layer, a normalization layer, a second LSTM hidden layer, a dropout layer, a first fully connected layer, a first activation layer, a second fully connected layer, a second activation layer, a third fully connected layer, and an output layer connected in sequence.

4. The method for identifying weak points in the voltage of an active power distribution network considering photovoltaic output according to claim 1, characterized in that, The specific process of step 3 is as follows: The following set of power flow equations is established for each node of the active distribution network: In the formula, P m Q is the injected active power at node m. m U is the reactive power injected into node m. m U is the voltage magnitude at node m. n Let θ be the voltage magnitude at node n. mn G represents the phase angle difference between node m and node n. mn Let B be the real admittance between node m and node n. mn Let n be the imaginary admittance between node m and node n, where n represents the node connected to node m. The above power flow equations are solved iteratively using the Jacobian matrix. During the solution process, the injected active power of the photovoltaic output nodes in the active distribution network is predicted using the photovoltaic output prediction model trained in step 2. The injected active power of other nodes is known, thus obtaining the voltage amplitude of each node in the active distribution network. The voltage deviation rate of each node is calculated based on the voltage amplitude. In the formula, V m U is the voltage deviation rate at node m. N Rated voltage; The voltage sensitivity of each node is calculated using the following formula: In the formula, S m Let be the voltage sensitivity of node m.

5. The method for identifying weak points in the voltage of an active power distribution network considering photovoltaic output according to claim 1, characterized in that, The specific process of step 4 is as follows: The degree of each node is calculated using the following formula: In the formula, D m Let be the degree of node m, f represent any other node in the active distribution network besides node m, and T represent the set of all nodes in the active distribution network. If node m and f are connected, then a mf =1, otherwise a mf =0; The betweenness centrality of each node is calculated using the following formula: In the formula, B m For the betweenness centrality of node m, g st σ represents the minimum number of impedance paths between any nodes s and t. st (m) represents the number of paths passing through node m in the minimum impedance path between any nodes s and t. The average path length of each node is calculated using the following formula: In the formula, L m Let be the average path length of node m, N be the number of all nodes in the active distribution network, and l(m,f) represent the shortest path length from node m to node f.

6. The method for identifying weak points in the voltage of an active power distribution network considering photovoltaic output according to claim 1, characterized in that, In step 5, the comprehensive voltage weakness index value of each node is calculated as follows: W=αV+βS+γD'+δB+εL In the formula, W is the comprehensive voltage weakness index value of the node, D'=1-D, V, S, D, B and L are the normalized index values ​​of the node's voltage deviation rate, voltage sensitivity, node degree, betweenness centrality and average path length, respectively; α, β, γ, δ and ε are the weight coefficients of each index value, which are calculated by the entropy weight method.

7. A computer device comprising a memory, a processor, and a computer program stored in the memory and capable of running on the processor, characterized in that, When the processor executes the computer program, it implements the steps of the active power distribution network voltage weak point identification method considering photovoltaic output as described in any one of claims 1 to 6.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by the processor, it implements the steps of the method for identifying weak points in the voltage of an active distribution network that takes into account photovoltaic output as described in any one of claims 1 to 6.

Images