Mountain area power distribution network grid division method considering disaster scene characteristics and related device

Abstract

The invention relates to the technical field of power distribution, in particular to a mountain area power distribution network grid division method considering disaster scene characteristics and a related device, and the method comprises the steps: obtaining map information of a target area; performing preliminary division on the map information of the target area by adopting a U-Net neural network to obtain a plurality of land parcels; obtaining road lattice characteristics of each land parcel, and quantifying the road lattice characteristics; according to the quantized road lattice features, an MST is generated by applying a graph theory; clustering the MST by using a Chameleon clustering method to obtain a refined grid, and completing the grid division of the power distribution network in the mountainous area; according to the method, efficient management of the mountain area power distribution network can be realized, the grid division condition under the extreme weather condition is clear, and disaster risks and power demands of different land parcels can be evaluated more accurately. The problems that in the prior art, due to the fact that influences of power equipment operation information, disaster scene characteristics and the like are not considered in power distribution grid division, grid division is not clear, and a reliable basis cannot be provided for operators are solved.

Description

Technical Field

[0001] The present invention relates to the technical field of electric power distribution, and in particular to a method for dividing a distribution network grid in a mountainous area taking into account characteristics of disaster scenarios and related devices. Background Art

[0002] With the continuous development of the economy, the gradual opening of the distribution market, the acceleration of urban and rural construction, and the increasing complexity of the structure and management of distribution networks in mountainous areas, in order to improve the quality of user services and meet the people's growing demand for electricity, the reasonable planning of distribution networks has become an important issue that power supply companies need to pay attention to. Traditional power grid planning methods have problems such as low accuracy and insufficient adaptability. The method of grid planning of distribution networks has emerged. The grid planning of distribution networks aims to achieve precise investment and efficient operation and maintenance through grid management, and improve the overall operation efficiency of the power grid and the user's electricity experience.

[0003] The gridding of distribution networks is a requirement for the refined operation and management of power grids. Usually, they are reasonably classified according to the load density, land use nature, development degree, administrative region, etc. of different regions and plots, so as to form several areas. The power grid is planned and managed based on the principle that the power supply of the areas does not overlap and is relatively independent. Existing research has conducted a lot of research on the gridding division method of distribution networks from the perspective of spatial load prediction and load characteristics. The research methods can be divided into qualitative divisions that rely on manual or experience and quantitative divisions based on quantitative characteristic attributes according to the analysis dimension.

[0004] Qualitative classification that relies on manual work or experience is based on classification according to land use nature, geographical location, administrative boundaries, etc., and then further adjusted and optimized based on regional load values. For example, Li Haitao et al. determined the target grid and transition grid of the planning grid based on the current grid structure, load forecast results and planning objectives of the planning grid (Li Haitao, Sun Bo, Wang Xuan. Distribution network grid planning method and its application [J]. Journal of Electric Power System and Automation, 2015, 27(S1): 33-37.); Wang Jing et al. proposed a three-layer structural model of "distribution grid, functional grid, and power grid" to adapt to the large-scale planning and construction of distribution networks, established a "three-level grid" planning system, stipulated different planning standards and planning methods for the three-level grids of distribution, function, and power consumption, and defined the entire process of "three-level grid" planning (Wang Jing. Research on Lianyungang "three-level grid" distribution network planning method [D]. Nanjing University of Science and Technology, 2018.); Zhang Lianyao divided the distribution grid based on regional geographical boundaries, traffic conditions, land use nature, etc., and made a long-term forecast of power supply load based on the divided grids. Based on the forecast results, he proposed a new The basic construction requirements of the additional substations and the line allocation plan provide a basis for the actual overall planning of the power grid in the experimental area (Zhang Lianyao. Power Planning and Design of the Guangdong-Guangxi Cooperation Zone Based on Grid [D]. Jilin University, 2020. DOI: 10.27162 / d.cnki.gjlin.2020.007640.); Ren Hongyu et al. considered the development of different regions and divided the grid into mature grids, developing grids, and uncertain grids, and then adjusted them according to the predicted load value and related municipal needs (Ren Hongyu, Wang Zhuding, Zhang Chao, et al. Optimization Model and Method for Grid Planning of High-voltage Distribution Network [J]. Automation of Electric Power Systems, 2019, 43(14): 151-158.). This method is usually based on the experience of experts and the understanding of regional characteristics to ensure the rationality and effectiveness of the division. However, this method has high professional requirements for the division personnel and a large workload. In practical applications, it has problems such as strong subjectivity and difficulty in quantification.

[0005] Most quantitative divisions based on quantitative characteristic attributes quantify the qualitative description of block characteristic attributes, and then cluster them according to the similarity between blocks. For example, Yang Fengren, based on the road distribution pattern of a certain place, considered the load and area differences between cells and used the main road and secondary road network for division (Yang Fengren. Medium-voltage distribution network grid planning method and automatic wiring algorithm based on load clustering and block division [D]. Chongqing University, 2016.); Fan Mingtian et al. proposed a method of applying the minimum spanning tree algorithm in graph theory to simplify the computational complexity of distribution network framework planning, taking the minimum total construction cost of the line as the objective function of planning optimization, and carrying out the framework planning method (Fan Mingtian, Zhang Zuping, Yue Zongbin translation. Distribution Network Planning and Design [M]. Beijing: China Electric Power Press, 1999.). However, in actual planning work, the construction cost of the line is related to the overall planning scheme and cannot be determined in advance, which will lead to the inability to carry out the minimum spanning tree algorithm with the line construction cost as the weight; In addition, there are also division methods that cluster with the minimum load difference and the minimum distance between blocks as the goal. However, most of the existing quantitative classification methods only classify from user-side dimensions such as minimal load difference, similar load density, and complementary load characteristics, ignoring the quantification of dimensions such as geographical geology, land use nature, and grid component information.

[0006] Therefore, the existing conventional substation grid division method only considers departmental planning information such as land use nature, administrative region, development degree, as well as load forecasting, load characteristics and other data, and does not involve various information that affects the operation of power equipment, making it difficult to achieve coupled optimization with the dispatching and control of the established distribution network substations. In addition, with the frequent occurrence of extreme weather, the existing grid division does not consider the characteristics of disaster scenarios, and the disaster attributes of the grid are unclear. There are blind spots in the actual grid management work, and it is impossible to provide guidance for the pre-distribution layout of personnel and materials in each grid of the distribution network before a disaster.

[0007] In the context of global warming and frequent extreme weather events, the grid division method that relies solely on load management can no longer meet the actual needs of safe operation of the power grid. In addition, the geographical environment, meteorological conditions, and grid structure of the mountain distribution network are relatively complex and special. Extreme rainstorms are one of the extreme weather conditions with a high frequency and severe losses in mountain distribution networks. They can easily induce secondary disasters such as mountain torrents, mudslides, and landslides. Therefore, it is urgent to consider the characteristics of different secondary disaster scenarios under extreme rainstorms, carry out more detailed grid division of the distribution network, and clarify the grids prone to various secondary disasters, so as to effectively guide the planning, operation and management of the power grid and identify the risks and vulnerabilities of the distribution network. Summary of the invention

[0008] In view of the problem in the prior art that the power distribution grid division is unclear and cannot provide a reliable basis for operators due to the lack of consideration of the influence of power equipment operation information, disaster scene characteristics, etc., the present invention provides a mountain distribution network grid division method and related devices that take disaster scene characteristics into consideration.

[0009] In order to achieve the above object, the present invention adopts the following technical solutions: The present invention provides a method for dividing a mountain distribution network grid taking into account disaster scenario characteristics, comprising: Get map information of the target area; The U-Net neural network is used to preliminarily divide the map information of the target area and obtain several plots; Obtain the road grid characteristics of each plot, and quantify the road grid characteristics to obtain quantified road grid characteristics; wherein the road grid characteristics of the plot include disaster scene characteristics, power plot attributes and grid component information characteristics; the disaster scene characteristics include terrain, elevation, slope, slope morphology, hydrological conditions and loose solid material stability coefficient; the power plot attributes include land use nature, load value and net load density; the grid component information characteristics include operating time, number of repair reports and material strength; Based on the quantified road grid features, graph theory is applied to generate MST; The Chameleon clustering method is used to cluster the MST, obtain the refined grid, and complete the grid division of the mountain distribution network.

[0010] Optionally, the map information of the target area includes road planning, water system distribution and power corridor map of the area to be divided.

[0011] Optionally, the method of using the U-Net neural network to preliminarily divide the map information of the target area to obtain a number of plots is: Confirm hard constraints and make it clear that channels cannot be crossed; After identifying the uncrossable channels, the map information of the target area is merged into an input data set to obtain training data, and the training data is annotated to create a label image; Perform data preprocessing on label images; The preprocessed labeled images are divided into training sets and validation sets, and the U-Net model is trained; The trained U-Net model is used to preliminarily divide the map information of the target area to obtain several plots.

[0012] Optionally, the method of confirming the hard constraint and making it clear that the channel cannot be crossed is: The hard constraints mentioned include municipal planning coordination, transportation facility restrictions, river basin boundaries, and power corridor protection; According to the hard constraints, the non-crossable channels are divided into three levels: Level 1 non-crossable corridors include administrative boundaries, railways, first-level highways, and river basins; Secondary non-crossable corridors include rail transit, expressways, secondary roads, major rivers, and power corridors; Level 3 uncrossable channels include level 3 roads and level 1 tributaries of major rivers.

[0013] Optionally, the method for obtaining the road grid characteristics of each land parcel includes: Obtaining the stability coefficient of loose solid materials:

[0014] in, is the stability factor of loose solid matter; is the internal friction angle of loose material; For cohesion; It is the density of loose material before soaking in water; It is the density of loose materials before and after soaking in water; is the density of water; is the density of the sand-laden water flow; is the thickness of saturated loose solid material; is the water depth on the slope; is the acceleration due to gravity; is the slope inclination; Obtaining the load value:

[0015] in, is the road load value; is the number of users in the road grid, is the maximum number of users in the road grid; For the region The electricity consumption of each user; Obtaining the net load density:

[0016] in, is the road load value; S is the road grid area; D is the net load density; Getting the running time:

[0017] in, It is the sum of the operating time of all grid components in the grid; is the number of grid elements in the grid; is the maximum number of grid elements in a grid; is the current time, Lugonetti The time when each grid component is put into use; Obtaining the number of repair reports:

[0018] in, It is the total number of warranty times for all power grid components in the grid; Lugonetti Number of maintenance times for power grid components; Material strength acquisition:

[0019] in, It is the sum of the material strengths of all grid components in the grid; q is the type of power grid element; Q It is the maximum value of the power grid element type; is the material strength of the grid element under the corresponding grid element type; b is the number of grid elements under the corresponding grid element type; B is the maximum number of grid elements under the corresponding grid element type.

[0020] Optionally, the method of generating MST by applying graph theory according to the quantified road grid features is: Obtain data on the main and branch roads for each plot of land; According to the data of traffic trunk roads and branch roads, a connecting circle is drawn with each road connection point as the center to obtain the mutually adjacent road grids of each road connection point; Calculating the centroids of mutually adjacent road grids at each road connection point respectively, and obtaining the centroid coordinates of mutually adjacent road grids at each road connection point; According to the centroid coordinates of the mutually adjacent road grids of each road connection point, the road grid connection relationship is established respectively to obtain a complete adjacency topology map; According to the quantified disaster scenario characteristics and the properties of the power-using plots, the weights of each edge of the adjacent topological graph are calculated; According to the weight of each edge of the adjacency topology graph, the Prim algorithm is used to generate the MST.

[0021] Optionally, the Chameleon clustering method is used to cluster the MST to obtain a refined grid, and the method for completing the grid division of the mountain distribution network is as follows: The hMetis algorithm is used to divide the MST into several unconnected subgraphs, and each subgraph is used as the initial subcluster for hierarchical clustering; The quantized information features of power grid components are used to calculate the similarity between the initial sub-clusters, and the initial sub-clusters with similarity higher than the preset threshold are merged until the number of preset result clusters is reached, and the refined grid is obtained to complete the grid division of the mountain distribution network.

[0022] The present invention provides a mountain distribution network grid division system taking into account disaster scenario characteristics, comprising: Map information acquisition module: used to obtain map information of the target area; Preliminary division module: used to use U-Net neural network to perform preliminary division of the map information of the target area to obtain several plots; Road grid feature acquisition and quantification module: used to obtain the road grid features of each plot, and quantify the road grid features to obtain quantified road grid features; wherein the road grid features of the plot include disaster scene features, power plot attributes and grid component information features; the disaster scene features include terrain, elevation, slope, slope morphology, hydrological conditions and loose solid material stability coefficient; the power plot attributes include land use nature, load value and net load density; the grid component information features include operating time, number of repair reports and material strength; MST generation module: used to generate MST based on quantified road grid features and apply graph theory; Refined grid module: It is used to cluster MST using the Chameleon clustering method to obtain refined grids and complete the grid division of the mountain distribution network.

[0023] A terminal device includes a memory, a processor, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps of the above-mentioned method for dividing a mountain distribution network grid taking into account disaster scenario characteristics are implemented.

[0024] A computer-readable storage medium stores a computer program, which, when executed by a processor, implements the steps of the above-mentioned method for dividing a mountain distribution network grid by considering disaster scenario characteristics.

[0025] Compared with the prior art, the present invention has the following beneficial effects: The present invention discloses a method for dividing a grid of a mountainous power distribution network taking into account disaster scene characteristics. The method first obtains map information of a target area; uses a U-Net neural network to preliminarily divide the map information of the target area to obtain a plurality of plots; by using the U-Net neural network to preliminarily divide the map information of the target area, the plot boundaries can be quickly and accurately extracted from the complex map information; the grid characteristics of each plot are obtained, and the grid characteristics are quantified to obtain quantified grid characteristics; wherein the grid characteristics of the plot include disaster scene characteristics, power plot attributes and power grid component information characteristics; the disaster scene characteristics include terrain, elevation, slope, slope morphology, hydrological conditions and loose solid material stability coefficient; the power plot attributes include land use properties, load value and net load density; the power grid component information characteristics include operating time, number of repair reports and material strength; the grid characteristics take into account the influence of disaster scene characteristics, power plot attributes and power grid component information characteristics; then scientific grid refinement is adopted, and a minimum spanning tree (Minimum Spanning Tree, MST), to ensure the scientificity and rationality of grid division; finally, the Chameleon clustering method is used to cluster MST, obtain refined grids, and complete the grid division of mountain distribution network; among them, the Chameleon clustering algorithm can find clusters of any shape, and has a good clustering effect on high-dimensional data, so that the refined grid is more in line with the actual situation of the plot, and the accuracy and practicality of the division are improved; the double-layer automatic grid division method proposed in the present invention can not only greatly reduce the manual workload, reduce the problem of strong subjectivity and high professional requirements of manual division, but also speed up the division speed, realize the organic combination of the two methods, realize the efficient management of mountain distribution network, clarify the grid division situation under extreme weather conditions, and more accurately evaluate the disaster risk and power demand of different plots, which is of great significance for formulating disaster response strategies, optimizing the layout of power facilities and improving the reliability of power supply. At the same time, the method can also provide a scientific basis for urban planning, land use and environmental protection and other fields to promote sustainable development.

[0026] The present invention provides a mountain distribution network grid division system that takes into account disaster scene characteristics. Through the settings of a map information acquisition module, a preliminary division module, a road grid feature acquisition and quantification module, an MST generation module and a refined grid module, the fine division of the mountain distribution network grid is achieved; wherein, through the map information acquisition module, the system can accurately obtain the map information of the target area, providing a solid foundation for subsequent division work; the preliminary division module uses a U-Net neural network to perform preliminary division of the map information of the target area. This deep learning method can efficiently process complex map information and preliminarily divide a number of plots; the road grid feature acquisition and quantification module not only takes into account the conventional power consumption plot attributes and power grid component information characteristics, but also specially introduces disaster scene characteristics, such as terrain, elevation, slope, slope morphology, etc. , hydrological conditions and loose solid material stability coefficient, etc. These characteristics are crucial for the planning and operation of mountain distribution networks, and can significantly improve the system's ability to identify and respond to disaster risks. Through the quantitative processing of road grid characteristics, the system can transform complex disaster scene characteristics, power plot attributes and grid component information characteristics into comparable and calculable numerical indicators, which is convenient for subsequent graph analysis and clustering processing; the MST generation module generates a minimum spanning tree based on the quantified road grid characteristics, which helps to minimize the total length or total cost of the network while maintaining network connectivity; the refined grid module clusters the MST using the Chameleon clustering method to further obtain a refined grid, which can consider the multi-level similarity between data, thereby obtaining a more reasonable and detailed grid division result. By comprehensively considering the disaster scene characteristics, power plot attributes and grid component information characteristics, the system can generate a distribution network grid division scheme that is more in line with the actual situation in the mountainous area, which helps to quickly locate and isolate the fault area when a disaster occurs, reduce the impact of disasters on the distribution network, and improve the safety and reliability of the distribution network.

[0027] The present invention also provides a terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein the processor implements the steps of the above method when executing the computer program. The device has a simple structure, low modification cost, and small resource occupation.

[0028] A computer-readable storage medium stores a computer program, wherein the computer program implements the steps of the above method when executed by a processor. The storage medium has good portability and strong versatility. BRIEF DESCRIPTION OF THE DRAWINGS

[0029] Figure 1 The present invention is a flow chart of a method for dividing a mountain distribution network grid by taking into account disaster scenario characteristics.

[0030] Figure 2 A schematic diagram of the construction of an adjacency topology in a method for dividing a mountain distribution network grid taking into account disaster scenario characteristics according to the present invention.

[0031] Figure 3 It is a simplified schematic diagram of the overall process of a method for dividing a mountain distribution network grid taking into account disaster scenario characteristics of the present invention.

[0032] Figure 4 This is a structural diagram of a mountain distribution network grid division system that takes into account disaster scenario characteristics according to the present invention. DETAILED DESCRIPTION

[0033] In order to enable those skilled in the art to better understand the scheme of the present invention, the technical scheme in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. 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 should fall within the scope of protection of the present invention.

[0034] It should be noted that the terms "first", "second", etc. in the specification and claims of the present invention and the above-mentioned drawings are used to distinguish similar objects, and are not necessarily used to describe a specific order or sequence. It should be understood that the data used in this way can be interchanged where appropriate, so that the embodiments of the present invention described herein can be implemented in an order other than those illustrated or described herein. In addition, the terms "including" and "having" and any variations thereof are intended to cover non-exclusive inclusions, for example, a process, method, system, product or device that includes a series of steps or units is not necessarily limited to those steps or units that are clearly listed, but may include other steps or units that are not clearly listed or inherent to these processes, methods, products or devices.

[0035] The present invention is further described in detail below in conjunction with specific embodiments, which are intended to explain the present invention rather than to limit it.

[0036] Compared with the urban distribution network targeted by traditional research, the geographical environment, meteorological conditions, and grid structure of the hybrid distribution network in mountainous areas are relatively special, and they are prone to secondary disasters such as debris flow, landslide, and waterlogging under extreme rainstorms. Towns with relatively concentrated populations are mostly concentrated in relatively flat areas near water bodies, which are similar to the geographical conditions of urban distribution networks. However, due to the incomplete urban and rural construction and defects in infrastructure, the drainage capacity and emergency repair capacity of urban and rural areas are poor, and most of the grid components are exposed to the ground. Under extreme rainstorms, they are very susceptible to waterlogging disasters and are at extremely high risk; while mountainous areas with relatively sparse populations have large undulating terrain and diverse soil conditions. Some areas with loose soil are very prone to disasters such as landslides and debris flow. The layout of the power grid lines is complex, and the height difference and direction of the lines are highly consistent with the terrain conditions, which can easily cause deformation of poles and towers and induce line shutdown. Therefore, in the context of global warming and frequent extreme weather events, the grid division method limited to load management can no longer meet the actual needs of safe operation of the power grid. In order to improve the efficiency of grid management, it is urgent to conduct feature analysis of different secondary disaster scenarios to provide a specific scenario information basis for grid division.

[0037] In practical applications, mountain distribution networks usually include two parts: towns and mountainous areas. The road network in the urban area is crisscrossed, and the number of roads increases rapidly as the road grade decreases. In the mountainous area, due to the relatively dispersed population distribution, there are usually only roads of higher grades. Grid division generally uses roads as grid boundaries. Most existing studies use a "bottom-up" division method to directly cluster the minimum grids in the study area and divide the grids with higher similarity into the same grid. However, this type of method often faces the problem of excessive information and complex calculations in practical applications. It is only applicable to small areas with small spans. In the grid division of mountain distribution networks with larger ranges and spans, its effect is not ideal.

[0038] See also Figures 1 to 3 In view of the above problems, the present invention discloses a method for dividing a mountain distribution network grid taking into account the characteristics of disaster scenarios, including: S1: Obtain map information of the target area; the map information of the target area is a map of road planning, water system distribution, power corridors, etc. of the target area with complete roads and accurate information.

[0039] S2: Use the U-Net neural network to preliminarily divide the map information of the target area and obtain several plots, specifically: U-Net is a convolutional neural network for image segmentation. It was first proposed by Olaf Ronneberger and his team in 2015 to address the high cost of manually annotating images and the limited number of labeled samples in biomedicine. The convolutional neural network (CNN) commonly used in the past performed well in feature extraction and classification, but still had limitations in accurately locating and segmenting complex biological images (such as cell nuclei or organs). U-Net has achieved remarkable results in cell and organ segmentation tasks, overcoming the challenges of difficult biomedical image labeling and limited data sets, allowing segmentation models to perform well on small-scale training data.

[0040] The core design of U-Net is to extract high-level contextual features through an encoder-decoder structure and retain image detail information through “skip connections” to achieve accurate pixel-level segmentation.

[0041] The role of the encoder is to extract the high-dimensional features of the image, while the decoder is used to gradually restore the spatial information and finally generate a segmentation map of the same size as the input image.

[0042] The encoder part consists of a series of convolutional layers, batch normalization, rectified linear unit (ReLU) activation function and max-pooling layer. The convolutional layer is used to extract features, and the max-pooling layer is used to gradually reduce the spatial resolution and retain key information. After being processed by the encoder, the spatial information of the image is reduced, while the semantic information is increased. Given an input image , each convolution operation of the encoder can be expressed as:

[0043] in, Indicates Feature maps of layers; is the encoding convolution kernel; is the encoding bias; is a nonlinear activation function (usually ReLU); is the convolution operation.

[0044] The decoder restores the spatial resolution of the image through up-sampling layers, and gradually restores the feature map to the original resolution through deconvolution. A "skip connection" is connected after each up-sampling step, so that the encoder's features are merged with the decoder's features at the corresponding position, thereby retaining more detailed information. The deconvolution operation of the decoder can be expressed as:

[0045] in, For decoder The feature map of the layer, To decode the deconvolution kernel, Decoding bias.

[0046] The skip connection directly inputs the feature maps of each layer in the encoder into the symmetric layer of the decoder, realizing the fusion of low-level features and high-level features. This mechanism prevents the network from losing detail information during the restoration process and improves segmentation accuracy.

[0047] The goal of U-Net is pixel-level segmentation, so its loss function is usually based on classification cross entropy (Cross-Entropy Loss) or Dice Similarity Coefficient (Dice Similarity Coefficient). Especially when used for segmentation tasks, Dice loss helps to deal with unbalanced data samples.

[0048] Cross entropy loss: given segmentation result and the true label , the cross entropy loss is defined as:

[0049] in, is the cross entropy loss; Dice loss function: The Dice loss function is based on the Dice similarity coefficient, which measures the overlap between the predicted area and the true area. The Dice similarity coefficient is defined as follows:

[0050] Among them, Dice represents the Dice similarity coefficient; The corresponding Dice loss function is:

[0051] in, is the Dice loss function, Dice represents the Dice similarity coefficient; Grid division is a systematic division of geographic space. In the grid planning of mountain distribution networks, planning facilities such as municipal areas, highways, railways, water distribution and power corridors often form obstacles to the combination of plots. When making preliminary plot divisions, these hard constraints must be considered to ensure the rationality and effectiveness of the division. Therefore, when using the U-Net neural network to make preliminary divisions of the map information of the target area, the hard constraints must be confirmed first to make it clear that the channels cannot be crossed; The hard constraints mainly include: Municipal planning coordination: Grid division must be consistent with municipal planning and cannot cross the boundaries of municipal districts and urban control detailed planning; Traffic facility restrictions: important traffic facilities (such as railways, roads, and rail transit) cannot be crossed to ensure the integrity and safety of the traffic facilities; River basin boundaries: Different river basins cannot cross each other, and rivers should serve as natural boundaries; Power corridor protection: All reserved and existing power corridors must remain uncrossed to protect the safe and stable operation of power facilities.

[0052] According to the above hard constraints, through actual analysis, the complexity of management and planning as well as the protection of important facilities are taken into consideration. For example, the first-level channel may require stricter restrictions, while the third-level channel is relatively flexible. The main non-crossable channels are divided into three levels. The specific method is as follows: Level 1 non-crossable corridors: involving key infrastructure and important borders, specifically administrative boundaries, railways, first-level highways and river basins; Secondary non-crossable corridors: involving secondary infrastructure and important transportation routes, with relatively flexible management requirements, specifically rail transit, expressways, secondary roads, major rivers, and power corridors; Level 3 non-crossable channels: involving secondary traffic and facilities, with more flexible management requirements, specifically level 3 roads and level 1 tributaries of major rivers; After identifying the channels that cannot be crossed, all layers of the map information of the target area are merged into an input data set to obtain training data. The training data is annotated and a label image is created. Channels of different colors are represented by specific values ​​as the basis for supervised training. Data preprocessing is performed on the label images. In order to improve the generalization ability of the model, data enhancement is performed on the training images, including rotation, translation, scaling, color transformation, etc. The pixel values ​​of the input images are normalized to the range of [0, 1] to facilitate U-Net model training. The preprocessed labeled images are divided into training sets and validation sets, and the U-Net model is trained. That is, the prepared images and labels are divided into training sets and validation sets. The Dice similarity coefficient is used as the optimization target to improve the segmentation effect. The U-Net model is trained with the training set, and the performance is monitored on the validation set. The hyperparameters (such as learning rate, batch size, etc.) are adjusted to improve the U-Net model effect. After the U-Net model training is completed, the U-Net model needs to be evaluated. The performance of the U-Net model can be evaluated on the test set, and the segmentation effect of the model is measured by the IoU (Intersection over Union) value. When calculating the IoU value, first determine the intersection area and union area of ​​the two bounding boxes and calculate their area ratio. The intersection area determines the coordinates of the intersection rectangle by finding the minimum and maximum values ​​of the two bounding boxes on the x-axis and y-axis, and then calculates its area. The area of ​​the union area is equal to the sum of the areas of the two bounding boxes minus the area of ​​the intersection area. Finally, the IoU value is the area of ​​the intersection area divided by the area of ​​the union area.

[0053] Finally, the trained U-Net model is used to preliminarily divide the map information of the target area to obtain several plots, that is, the trained U-Net model is used to infer the new map information, generate a segmentation result map, and obtain several plots.

[0054] S3: Obtain the road grid characteristics of each plot, and quantify the road grid characteristics to obtain quantified road grid characteristics; wherein the road grid characteristics of the plot include disaster scene characteristics, power plot attributes and grid component information characteristics; the disaster scene characteristics include terrain, elevation, slope, slope morphology, hydrological conditions and loose solid material stability coefficient; the power plot attributes include land use nature, load value and net load density; the grid component information characteristics include operating time, number of repair reports and material strength, specifically: Based on the analysis of the mechanism of different secondary disaster scenarios under heavy rain, the characteristics of disaster scenarios are refined as follows: Terrain: When the terrain is plain (generally below 200 meters above sea level), the value is 0.5-1.0; when it is hilly (absolute height within 500 meters, relative height not exceeding 200 meters), the value is 1.0-1.5; when it is mountainous (highland with an altitude of more than 500 meters), the value is 1.5-2.0.

[0055] Elevation: Since the terrain of the distribution network in mountainous areas is undulating and the elevation varies from place to place, the elevation of the highest point in the area is taken as the elevation value; Slope: When the slope is between 20° and 40°, the adjustment coefficient is 1.0; when the slope is greater than 40°, the adjustment coefficient is 1.0 to 1.5; when the slope is less than 20°, the adjustment coefficient is 0.5 to 1.0; Slope shape: Taking the straight slope as the benchmark, the slope value is 1.0-1.5 for convex slopes and 0.5-1.0 for concave slopes; Hydrological conditions: Based on the situation where there is basically no surface runoff and the groundwater is buried very deep, the value is 1.0 to 2.0 according to the changes in surface runoff and groundwater burial depth; Stability factor of loose solid matter:

[0056] in, is the stability factor of loose solid matter; is the internal friction angle of loose material; For cohesion; It is the density of loose material before soaking in water; It is the density of loose materials before and after soaking in water; is the density of water; is the density of the sand-laden water flow; is the thickness of saturated loose solid material; is the water depth on the slope; is the acceleration due to gravity; is the slope inclination; The properties of the electricity plots are summarized as follows: Land use nature: the land use nature of the location with the maximum load value in the road grid, which is a categorical variable; Load value: The total electricity consumption of all users in the grid, that is, the load value of the grid, as shown in the following formula, which is a numerical variable:

[0057] in, is the road load value; is the number of users in the road grid, is the maximum number of users in the road grid; For the region The electricity consumption of each user; Net load density: It is defined as the ratio of the total grid load to the grid area, as shown in the following formula, which is a numerical variable:

[0058] in, is the road load value; S is the road grid area; D is the net load density; The information characteristics of power grid components are summarized as follows: Running time: The total running time of all grid components in the grid, as shown in the following formula, which is a numerical variable:

[0059] in, It is the sum of the operating time of all grid components in the grid; is the number of grid elements in the grid; is the maximum number of grid elements in a grid; is the current time, Lugonetti The time when each grid component is put into use; Repair times: The total number of repair times for all grid components in the grid, as shown in the following formula, which is a numerical variable:

[0060] in, It is the total number of warranty times for all power grid components in the grid; Lugonetti Number of maintenance times for power grid components; Material strength acquisition:

[0061] in, It is the sum of the material strengths of all grid components in the grid; q is the type of power grid element; Q It is the maximum value of the power grid element type; is the material strength of the grid element under the corresponding grid element type; b is the number of grid elements under the corresponding grid element type; B is the maximum number of grid elements under the corresponding grid element type.

[0062] According to the above part, the geographical features of the disaster scene, the attributes of the power plot, the grid components and other features of each road grid are quantitatively processed. By quantifying the road grid features, the surface features are converted into point features, providing a data basis for the subsequent graph theory and Chameleon clustering method. In order to facilitate the subsequent similarity calculation of road grid features, the present invention first replaces each feature variable type with a unified symbol representation, and the classification variable is , the numerical variable is The specific meaning and function of the quantified road grid characteristics are shown in the following table:

[0063] S4: Based on the quantified road grid features, graph theory is used to generate MST, specifically: Obtain data on the main and branch roads of each plot of land, that is, collect data on the main and branch roads of the city, including intersection coordinates, length and other information. Ensure the accuracy and completeness of the data for subsequent analysis; According to the data of traffic trunk roads and branch roads, draw connecting circles with each road connection point as the center of the circle, and obtain the mutually adjacent road grids of each road connection point, that is, for each pair of road connection points (intersections), draw a circle with a radius of 100 meters with the coordinates of the road connection point as the center of the circle; check which roads intersect with the circle, and any road section intersecting with the circle is considered to be mutually adjacent. Spatial analysis tools (such as Geographic Information System (GIS) software) can be used to perform this operation, identify and record all intersecting road sections, and obtain the mutually adjacent road grids of each road connection point; Calculate the centroid of the mutually adjacent grids of each road connection point respectively, and obtain the centroid coordinates of the mutually adjacent grids of each road connection point, that is, for each mutually adjacent grid determined in the previous step, calculate the centroid of each road section, which is the geometric center of the road section, usually obtained by calculating the average value of the coordinates of the end points of the road section, and store the centroid coordinates of each road section for use in subsequent connection operations; According to the centroid coordinates of the mutually adjacent grids of each road connection point, the grid connection relationship is established respectively to obtain a complete adjacency topology map, that is, starting from the centroid of any grid, traverse the centroids of all adjacent grids clockwise, record the order of these centroids, until returning to the starting point. The closed path formed by this process represents the connection relationship of the grids; repeat the above operation for all intersections in the city to ensure that the connection relationship of each intersection is recorded, so as to establish a complete adjacency topology map; see Figure 2 , a schematic diagram of the adjacency topology constructed for a certain plot.

[0064] According to the quantified disaster scene characteristics and power plot attributes, the weight of each edge of the adjacent topological graph is calculated, that is, for each edge of the adjacent topological composition, the similarity of the disaster scene characteristics and power plot attribute information of its adjacent road grid is calculated as the weight of each edge. Road grid similarity calculation can be calculated by the Euclidean distance of multi-dimensional information of adjacent road grids. Considering the actual situation and application area, different weights can be assigned to each information, thereby reflecting the importance of different information to the similarity calculation. The specific calculation method is as follows:

[0065] in, For two adjacent grids With Luger The similarity between is the number of numerical variables in the road disaster scene characteristics and power consumption plot attributes (the value range is 1 to 3), is the number of categorical variables in the disaster scene characteristics and electricity plot attributes (ranging from 1 to 6); For Luger The first of the Luger characteristics n The value of a numeric variable; For Luger The first of the Luger characteristics n The value of a numeric variable; For Luger The first of the Luger characteristics The value of a categorical variable; For Luger The first of the Luger characteristics The value of a categorical variable; The first n The maximum value of a numeric variable; The first n The minimum value of a numeric variable; is the information weight of the numerical variable and is the information weight of categorical variables; Adjacent road grid and The variable type is a numeric variable. n The intermediate variable of the similarity of the road characteristics; For adjacent road grid and The variable type is categorical variable. m The intermediate variable of the similarity of the road characteristics; According to the weight of each edge of the adjacent topological graph, the Prim algorithm is used to generate the MST, that is, the weight of each edge is calculated through the above process. The smaller the weight, the higher the similarity of its adjacent grids. When the grid is divided, the grid can be merged and the edge can be deleted. Therefore, the Prim algorithm is considered to construct the MST, so as to simplify the topological structure and provide a basis for subsequent grid aggregation. Specifically: Randomly select a vertex from the adjacency topology graph as the starting point, such as vertex A; initialize an empty set MST to store the edges of the minimum spanning tree, and maintain a set selected to store the selected vertices, which initially only contains the starting vertex A. Use a priority queue (or minimum heap) to store the edges from the selected vertices to the unselected vertices, and sort them according to the weight of the edges, find all the edges adjacent to the starting vertex A, and add these edges to the priority queue. The priority queue stores the weight of the edge and the connected vertices, such as (weight, vertex 1, vertex 2).

[0066] Then, extract the edge with the smallest weight from the priority queue, recorded as (weight, u, v). Here, u is the selected vertex and v is the unselected vertex. Check whether vertex v is already in the set selected. If not, it means that this edge is valid and can be added to the minimum spanning tree. Update the priority queue, find all unselected edges adjacent to the newly added vertex v, and add these edges to the priority queue; if it is already in selected, discard this edge and continue to extract the next edge from the priority queue. Repeat the above operations. When all vertices are selected, that is, the selected set contains all vertices in the adjacent topology graph, the algorithm ends and outputs the MST.

[0067] S5: Use the Chameleon clustering method to cluster MST, obtain refined grids, and complete the grid division of the mountain distribution network; the above part divides the study area according to information such as disaster scene characteristics and power-consuming land attributes. These two types of information change less over time and can be considered to be basically unnecessary to update. However, in the grid processing of the distribution network, multidimensional data reflecting the operation of power equipment of power users, such as power user load records, repair records, and operating years, need to be considered. These information change relatively greatly over time and need to be updated in time and re-divided into grids. If the recent load levels of multiple power users at a certain location are too high and repairs are frequent, or the power grid components at a certain location have been in operation for a long time and have low material strength, they can be classified into one grid to facilitate the dispatch of maintenance personnel to focus on them. Therefore, this method combines the information characteristics of power grid components such as operating time, number of repairs, and material strength, and reasonably clusters them based on the improved Chameleon method, thereby effectively improving the management and maintenance efficiency of the distribution network. Specifically: The hMetis algorithm is used to divide the MST into several unconnected subgraphs, and each subgraph is used as the initial subcluster for hierarchical clustering. The MST obtained in the previous part is regarded as a special graph, and the hMetis algorithm is used to divide the MST into many unconnected subgraphs. Since the MST is an acyclic graph, its rationality must be determined when setting the parameters for dividing the subgraphs. Each subgraph is used as the initial subcluster for hierarchical clustering. The similarity between the initial sub-clusters is calculated by using the quantized information characteristics of the power grid components, and the initial sub-clusters with similarity higher than the preset threshold are merged until the number of the preset result clusters is reached, and the refined grid is obtained to complete the grid division of the mountain distribution network. That is, the relative interconnectivity (Relative Interconnectivity, RI) and relative closeness (Relative Closeness, RC) between the two clusters are considered to calculate the similarity between the two clusters. The calculation method is as follows; when the similarity function value is higher than the preset threshold, the two sub-clusters are merged, and the above merging process is repeated until the number of the preset result clusters is reached, and the refined grid is obtained to complete the grid division of the mountain distribution network:

[0068] in, Cluster With cluster The relative interconnectedness between Cluster With cluster The relative similarity between To connect clusters With cluster All edges between; Cluster The sum of the weights of all internal edges; Cluster The sum of the weights of all internal edges; Cluster The total number of edges contained in the interior; Cluster The total number of edges contained in the interior; Cluster With cluster The average weight of the connecting line, Cluster The average weight of the internal connecting lines; Cluster The average weight of the internal connecting lines; is the similarity function between two clusters, is the weight of the metric, which can be changed according to actual needs; in, The calculation method of is the same as the weight calculation of each edge in the previous part of the topology. However, since the information characteristics of the actual power grid components will change over time (for example, as the city develops and time passes, the operation time and repair frequency of the power grid components may increase, and the material strength may also change due to the replacement of power grid components, etc.), only the operation time is involved in the weight calculation. , Number of repair reports and material strength ;

[0069] in, Cluster With cluster The variable type is a numeric variable. The intermediate variable of the similarity of the road features, Cluster No. The value of the feature variable, Cluster No. The value of the feature variable, The first n The maximum value of a numeric variable; The first n The minimum value of a numeric variable; is the information weight of the numerical variable; contains multiple grids. The value of the characteristic variable It is necessary to sum the values ​​of all the grids in the cluster, as shown in the following formula:

[0070] in, Cluster Middle Luger's The value of a feature variable.

[0071] See also Figure 4 The present invention also provides a mountain distribution network grid division system considering disaster scene characteristics, including: Map information acquisition module: used to obtain map information of the target area; Preliminary division module: used to use U-Net neural network to perform preliminary division of the map information of the target area to obtain several plots; Road grid feature acquisition and quantification module: used to obtain the road grid features of each plot, and quantify the road grid features to obtain quantified road grid features; wherein the road grid features of the plot include disaster scene features, power plot attributes and grid component information features; the disaster scene features include terrain, elevation, slope, slope morphology, hydrological conditions and loose solid material stability coefficient; the power plot attributes include land use nature, load value and net load density; the grid component information features include operating time, number of repair reports and material strength; MST generation module: used to generate MST based on quantified road grid features and apply graph theory; Refined grid module: It is used to cluster MST using the Chameleon clustering method to obtain refined grids and complete the grid division of the mountain distribution network.

[0072] The present invention provides a terminal device including: a processor, a memory, and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the steps in the above-mentioned method embodiments are implemented. Alternatively, when the processor executes the computer program, the functions of the modules / units in the above-mentioned device embodiments are implemented.

[0073] The computer program may be divided into one or more modules / units, and the one or more modules / units are stored in the memory and executed by the processor to accomplish the present invention.

[0074] The terminal device may be a computing device such as a desktop computer, a notebook, a palmtop computer, a cloud server, etc. The terminal device may include, but is not limited to, a processor and a memory.

[0075] The processor may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA) or other programmable logic devices, discrete gate or transistor logic devices, discrete hardware components, etc.

[0076] The memory may be used to store the computer programs and / or modules, and the processor implements various functions of the terminal device by running or executing the computer programs and / or modules stored in the memory and calling the data stored in the memory.

[0077] If the module / unit integrated in the terminal device is implemented in the form of a software functional unit and sold or used as an independent product, it can be stored in a computer-readable storage medium. Based on this understanding, the present invention implements all or part of the processes in the above-mentioned embodiment method, and can also be completed by instructing the relevant hardware through a computer program. The computer program can be stored in a computer-readable storage medium. When the computer program is executed by the processor, the steps of the above-mentioned various method embodiments can be implemented. Among them, the computer program includes computer program code, and the computer program code can be in source code form, object code form, executable file or some intermediate form. The computer-readable medium may include: any entity or device capable of carrying the computer program code, recording medium, U disk, mobile hard disk, magnetic disk, optical disk, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signal, telecommunication signal and software distribution medium. It should be noted that the content contained in the computer-readable medium can be appropriately increased or decreased according to the requirements of legislation and patent practice in the jurisdiction. For example, in some jurisdictions, according to legislation and patent practice, computer-readable media do not include electric carrier signals and telecommunication signals.

[0078] In summary, the present invention provides a method and related devices for grid division of mountain distribution networks that considers disaster scene characteristics. By analyzing the mechanism of secondary disasters after extreme rainstorms, the occurrence characteristics of different disaster scenes are extracted, and multi-dimensional characteristics such as disaster scene characteristics, power consumption plot attributes, and grid component information characteristics are quantitatively processed. Then, a method for extracting grid characteristics of mountain distribution networks that considers the impact of secondary disasters is proposed. The method is divided into two layers of grids. The first layer considers the hard constraints of grid division and uses the U-Net neural network for preliminary division. The second layer uses graph theory and Chameleon clustering methods to cluster the minimum grids of each plot preliminarily divided by the first layer. Through this comprehensive grid division method, more efficient management of the distribution network can be achieved, and the grid division situation under extreme weather conditions can be clarified, laying a solid foundation for responding to the challenges brought by future climate change.

[0079] The above description is only a preferred embodiment of the present invention and is not intended to impose any limitation on the technical solution of the present invention. Those skilled in the art should understand that, without departing from the spirit and principles of the present invention, the technical solution can also be subjected to several simple modifications and substitutions, and these modifications and substitutions are also within the scope of protection covered by the claims.

Claims

1. A method for grid division of mountain distribution network considering disaster scene characteristics, characterized in that: include: Get map information of the target area; The U-Net neural network is used to preliminarily divide the map information of the target area and obtain several plots; Obtain the road grid characteristics of each plot, and quantify the road grid characteristics to obtain quantified road grid characteristics; wherein the road grid characteristics of the plot include disaster scene characteristics, power plot attributes and grid component information characteristics; the disaster scene characteristics include terrain, elevation, slope, slope morphology, hydrological conditions and loose solid material stability coefficient; the power plot attributes include land use nature, load value and net load density; the grid component information characteristics include operating time, number of repair reports and material strength; Based on the quantified road grid features, graph theory is applied to generate MST; The Chameleon clustering method is used to cluster the MST, obtain the refined grid, and complete the grid division of the mountain distribution network.

2. The method for grid division of mountain distribution network considering disaster scene characteristics according to claim 1 is characterized in that: The map information of the target area includes road planning, water system distribution and power corridor map of the area to be divided.

3. The method for grid division of mountain distribution network considering disaster scene characteristics according to claim 1 is characterized in that: The method of using the U-Net neural network to preliminarily divide the map information of the target area to obtain several plots is as follows: Confirm hard constraints and make it clear that channels cannot be crossed; After identifying the uncrossable channels, the map information of the target area is merged into an input data set to obtain training data, and the training data is annotated to create a label image; Perform data preprocessing on label images; The preprocessed labeled images are divided into training sets and validation sets, and the U-Net model is trained; The trained U-Net model is used to preliminarily divide the map information of the target area to obtain several plots.

4. The method for dividing the grid of the mountain distribution network considering the characteristics of the disaster scene according to claim 3 is characterized in that: The method of confirming the hard constraints and making it clear that the channel cannot be crossed is: The hard constraints mentioned include municipal planning coordination, transportation facility restrictions, river basin boundaries, and power corridor protection; According to the hard constraints, the non-crossable channels are divided into three levels: Level 1 non-crossable corridors include administrative boundaries, railways, first-level highways, and river basins; Secondary non-crossable corridors include rail transit, expressways, secondary roads, major rivers, and power corridors; Level 3 uncrossable channels include level 3 roads and level 1 tributaries of major rivers.

5. The method for grid division of mountain distribution network considering disaster scene characteristics according to claim 1 is characterized in that: The method for obtaining the road grid characteristics of each plot includes: Obtaining the stability coefficient of loose solid materials: in, is the stability factor of loose solid matter; is the internal friction angle of loose material; For cohesion; It is the density of loose material before soaking in water; It is the density of loose materials before and after soaking in water; is the density of water; is the density of the sand-laden water flow; is the thickness of saturated loose solid material; is the water depth on the slope; is the acceleration due to gravity; is the slope inclination; Obtaining the load value: in, is the road load value; is the number of users in the road grid, is the maximum number of users in the road grid; For the region The electricity consumption of each user; Obtaining the net load density: in, is the road load value; S is the road grid area; D is the net load density; Getting the running time: in, It is the sum of the operating time of all grid components in the grid; is the number of grid elements in the grid; is the maximum number of grid elements in a grid; is the current time, Lugonetti The time when each grid component is put into use; Obtaining the number of repair reports: in, It is the total number of warranty times for all power grid components in the grid; Lugonetti Number of maintenance times for power grid components; Material strength acquisition: in, It is the sum of the material strengths of all grid components in the grid; q is the type of power grid element; Q It is the maximum value of the power grid element type; is the material strength of the grid element under the corresponding grid element type; b is the number of grid elements under the corresponding grid element type; B is the maximum number of grid elements under the corresponding grid element type.

6. The method for grid division of mountain distribution network considering disaster scene characteristics according to claim 1 is characterized in that: The method of generating MST by applying graph theory based on the quantified road grid features is: Obtain data on the main and branch roads for each plot of land; According to the data of traffic trunk roads and branch roads, a connecting circle is drawn with each road connection point as the center to obtain the mutually adjacent road grids of each road connection point; Calculating the centroids of mutually adjacent road grids at each road connection point respectively, and obtaining the centroid coordinates of mutually adjacent road grids at each road connection point; According to the centroid coordinates of the mutually adjacent road grids of each road connection point, the road grid connection relationship is established respectively to obtain a complete adjacency topology map; According to the quantified disaster scenario characteristics and the properties of the power-using plots, the weights of each edge of the adjacent topological graph are calculated; According to the weight of each edge of the adjacency topology graph, the Prim algorithm is used to generate the MST.

7. The method for grid division of mountain distribution network considering disaster scene characteristics according to claim 1 is characterized in that: Using the Chameleon clustering method, MST is clustered to obtain a refined grid, and the method for completing the grid division of the mountain distribution network is as follows: The hMetis algorithm is used to divide the MST into several unconnected subgraphs, and each subgraph is used as the initial subcluster for hierarchical clustering; The quantized information features of power grid components are used to calculate the similarity between the initial sub-clusters, and the initial sub-clusters with similarity higher than the preset threshold are merged until the number of preset result clusters is reached, and the refined grid is obtained to complete the grid division of the mountain distribution network.

8. A mountain distribution network grid division system considering disaster scenario characteristics, characterized in that: include: Map information acquisition module: used to obtain map information of the target area; Preliminary division module: used to use U-Net neural network to perform preliminary division of the map information of the target area to obtain several plots; Road grid feature acquisition and quantification module: used to obtain the road grid features of each plot, and quantify the road grid features to obtain quantified road grid features; wherein the road grid features of the plot include disaster scene features, power plot attributes and grid component information features; the disaster scene features include terrain, elevation, slope, slope morphology, hydrological conditions and loose solid material stability coefficient; the power plot attributes include land use nature, load value and net load density; the grid component information features include operating time, number of repair reports and material strength; MST generation module: used to generate MST based on quantified road grid features and apply graph theory; Refined grid module: It is used to cluster MST using the Chameleon clustering method to obtain refined grids and complete the grid division of the mountain distribution network.

9. A terminal device, comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, characterized in that: When the processor executes the computer program, the steps of the method for dividing a mountain distribution network grid taking into account disaster scenario characteristics as described in any one of claims 1 to 7 are implemented.

10. A computer-readable storage medium storing a computer program, characterized in that: When the computer program is executed by a processor, the steps of the method for dividing a mountain distribution network grid taking into account disaster scenario characteristics as described in any one of claims 1 to 7 are implemented.

Images