Method and device for optimizing layout of wind speed monitoring points along high-speed rail and storage medium

Abstract

The invention relates to a layout optimization method and device for wind speed monitoring points along a high-speed rail and a storage medium, and the method comprises the steps: obtaining the wind speed data of each monitoring point at an initial layout position, and constructing a multi-dimensional correlation index set according to the wind speed data between every two monitoring points; and carrying out weight distribution on the index set to obtain a weight distribution result, and calculating an association weight between each pair of monitoring points according to the weight distribution result. And building an initial association network of the monitoring points based on the weight distribution result and the association weight. In all connections of the initial association network, determining that the connections with association weights lower than a preset threshold are weak connections and removing the weak connections to form a corrected target association network; and adjusting the initial layout position of the monitoring point to the optimized layout position based on the target association network. According to the method and the device, the problems of overlarge alarm range, unreasonable resource allocation and insufficient high-risk area identification caused by local view angle layout of the wind speed monitoring points along the high-speed rail are solved, and the early warning precision and efficiency are improved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of high-speed rail layout optimization, in particular to a high-speed rail along the line wind speed monitoring point layout optimization method, device and storage medium. BACKGROUND

[0002] With the rapid expansion of high-speed railway network, the train running speed is continuously improved, and the high-speed rail operation safety is facing increasingly complex natural environment challenges, in which strong wind weather is one of the important factors affecting the safe operation of high-speed rail. Therefore, accurately and timely monitoring the wind speed change along the high-speed rail has important significance for ensuring the safe operation of high-speed rail and improving the operation efficiency.

[0003] At present, a wind speed monitoring point is set every 10 kilometers along the high-speed rail, mainly according to the terrain conditions, meteorological data and the experience of designers to arrange the interval. This method mainly depends on the interval arrangement, and does not fully consider the meteorological elements and terrain conditions, resulting in insufficient monitoring coverage of the high wind speed section and excessive data collection redundancy of the small wind speed section. This arrangement makes the monitoring data inaccurate, and is prone to problems such as strong wind shaking the train, long monitoring point alarm speed limit range, and some monitoring points not alarming all year round. Due to unreasonable arrangement of monitoring points, when strong wind occurs, multiple adjacent monitoring points along the line will trigger an alarm at the same time, resulting in a long alarm range and affecting the efficiency of operation. In addition, the existing method cannot effectively identify the weak points along the whole high-speed rail line, so that the early warning system cannot accurately capture the extreme wind speed.

[0004] At present, there is no effective solution to the problem of excessive alarm range, unreasonable resource allocation and insufficient identification of high-risk areas of wind speed monitoring points along the high-speed rail caused by local perspective layout in the related technology. SUMMARY

[0005] The embodiments of the present application provide a high-speed rail along the line wind speed monitoring point layout optimization method, device, system, electronic device and storage medium, to at least solve the problem of excessive alarm range, unreasonable resource allocation and insufficient identification of high-risk areas of wind speed monitoring points along the high-speed rail caused by local perspective layout in the related technology.

[0006] In a first aspect, the embodiments of the present application provide a high-speed rail along the line wind speed monitoring point layout optimization method, comprising:

[0007] Obtaining the wind speed data of each monitoring point in the initial layout position; based on the wind speed data between each two monitoring points, generating a multi-dimensional correlation index set;

[0008] generate a weight distribution result based on the multi-dimensional correlation index set, and calculate an association weight between each pair of the monitoring points;

[0009] In all connections of the initial association network, the connection with an association weight lower than a preset weight threshold is confirmed as a weak connection, the weak connection is deleted, and a modified target association network is formed; and the initial layout position of the monitoring point is adjusted to an optimized layout position based on the target association network.

[0010] In some embodiments, the multi-dimensional correlation index set is generated based on the wind speed data between each pair of the monitoring points, including:

[0011] The linear correlation coefficient, the nonlinear rank correlation coefficient, the similarity, and the distance metric index are calculated based on the wind speed data between each pair of the monitoring points, and the multi-dimensional correlation index set is generated based on the linear correlation coefficient, the nonlinear rank correlation coefficient, the similarity, and the distance metric index.

[0012] In some embodiments, the weight distribution result is generated based on the multi-dimensional correlation index set, including:

[0013] The multi-dimensional correlation index set is standardized, the information entropy and the information utility value corresponding to each index are calculated, and the weight distribution result of each index is determined based on the information utility value.

[0014] In some embodiments, the preset weight threshold is determined according to an obtained network structure index, and the network structure index includes a clustering coefficient, a network efficiency, an average degree, and a number of connected components of the initial association network.

[0015] In some embodiments, the initial layout position of the monitoring point is adjusted to the optimized layout position based on the target association network, including:

[0016] The monitoring point priority ranking result is calculated for each monitoring point of the target association network, and the initial layout position of the monitoring point is adjusted to the optimized layout position based on the target association network and the monitoring point priority ranking result.

[0017] In some embodiments, the monitoring point priority ranking result is calculated for each monitoring point of the target association network, including:

[0018] The multi-dimensional node importance evaluation index is calculated for each monitoring point of the target association network.

[0019] Based on the multi-dimensional node importance evaluation index, a node index actual value is calculated;

[0020] By calculating a total deviation value of the node index actual value and a preset index ideal value, a monitoring point priority ranking result is determined.

[0021] In some embodiments, the calculation of the node index actual value based on the multi-dimensional node importance evaluation index comprises:

[0022] Based on the wind speed data, a coefficient of variation is calculated; based on the multi-dimensional correlation index set and the coefficient of variation, an importance index weight for the multi-dimensional node importance evaluation index is determined;

[0023] Based on the multi-dimensional node importance evaluation index and the importance index weight, a node index actual value is calculated.

[0024] In some embodiments, the adjustment of the initial layout position of the monitoring point to an optimized layout position based on the target correlation network and the monitoring point priority ranking result comprises:

[0025] The target correlation network is divided into modules to obtain a correlation network module, and the correlation network module is matched with the terrain feature based on the obtained terrain feature of the monitoring point to obtain a terrain feature verification result;

[0026] Based on the target correlation network, the monitoring point priority ranking result, and the terrain feature verification result, the initial layout position of the monitoring point is adjusted to an optimized layout position.

[0027] In a second aspect, the embodiments of the present application provide a high-speed rail line wind speed monitoring point layout optimization device, comprising:

[0028] A data acquisition module is configured to acquire wind speed data of each monitoring point in an initial layout position; and based on the wind speed data between each pair of monitoring points, a multi-dimensional correlation index set is generated.

[0029] A network construction module is configured to perform weight allocation for each index based on the multi-dimensional correlation index set, generate a weight allocation result, and calculate a correlation weight between each pair of monitoring points; and based on the weight allocation result and the correlation weight, an initial correlation network for the monitoring points is constructed.

[0030] a network correction module, configured to: in all connections of the initial association network, determine a connection with an association weight lower than a preset weight threshold as a weak connection, and delete the weak connection, to form a corrected target association network; and based on the target association network, adjust the initial layout position of the monitoring point to an optimized layout position.

[0031] In a third aspect, an embodiment of the present application provides a storage medium having a computer program stored thereon, the program being executed by a processor to implement the high-speed rail line wind speed monitoring point layout optimization method according to the first aspect.

[0032] Compared with the related art, the high-speed rail line wind speed monitoring point layout optimization method, device and storage medium provided by the embodiments of the present application optimize the layout of the monitoring point through global network analysis and multi-index decision, solve the problems of excessively large alarm range, unreasonable resource allocation and insufficient identification of high-risk areas caused by the local perspective layout of the high-speed rail line wind speed monitoring point, and improve the early warning accuracy and efficiency.

[0033] Details of one or more embodiments of the present application are presented in the following drawings and description to make other features, objects and advantages of the present application more apparent. BRIEF DESCRIPTION OF DRAWINGS

[0034] The drawings described herein are intended to provide further understanding of the present application, form a part of the present application, and the illustrative embodiments of the present application and their descriptions are used to explain the present application and do not constitute improper limitations on the present application. In the drawings:

[0035] Figure 1 is a hardware structure block diagram of a terminal of the high-speed rail line wind speed monitoring point layout optimization method according to an embodiment of the present application;

[0036] Figure 2 is a flowchart of the high-speed rail line wind speed monitoring point layout optimization method according to an embodiment of the present application;

[0037] Figure 3 is a preferred flowchart of the high-speed rail line wind speed monitoring point layout optimization method according to an embodiment of the present application;

[0038] Figure 4 is a schematic diagram of wind speed monitoring point distribution according to the high-speed rail line wind speed monitoring point layout optimization method according to an embodiment of the present application;

[0039] Figure 5 is a schematic diagram of cleaned wind speed data according to the high-speed rail line wind speed monitoring point layout optimization method according to an embodiment of the present application;

[0040] Figure 6 is a sensitivity analysis schematic diagram according to the high-speed rail line wind speed monitoring point layout optimization method according to an embodiment of the present application;

[0041] Figure 7 is a schematic diagram of a surrounding terrain of a monitoring point according to the high-speed rail line wind speed monitoring point layout optimization method of the embodiments of the present application;

[0042] Figure 8 is a schematic diagram of community attribution of a monitoring point according to the high-speed rail line wind speed monitoring point layout optimization method of the embodiments of the present application;

[0043] Figure 9 is a schematic diagram of high-speed rail line wind speed and direction instrument arrangement according to the high-speed rail line wind speed monitoring point layout optimization method of the embodiments of the present application;

[0044] Figure 10 is a structural block diagram of the high-speed rail line wind speed monitoring point layout optimization device according to the embodiments of the present application. DETAILED DESCRIPTION

[0045] In order to make the objectives, technical solutions and advantages of the present application clearer, the present application is described and explained below in combination with the drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present application and should not be used to limit the present application. Based on the embodiments provided in the present application, all other embodiments obtained by those of ordinary skill in the art without making creative efforts fall within the scope of protection of the present application. In addition, it should be understood that although the efforts made in this development process can be complex and lengthy, some design, manufacture or production changes made on the basis of the technical content disclosed in the present application by those of ordinary skill in the art related to the content disclosed in the present application are only routine technical means and should not be understood as insufficient disclosure of the present application.

[0046] In the present application, the phrase "embodiments" means that the specific features, structures or characteristics described in combination with the embodiments can be included in at least one embodiment of the present application. The appearance of this phrase at various places in the specification does not necessarily mean the same embodiment, nor is it an independent or alternative embodiment to other embodiments. Those of ordinary skill in the art explicitly and implicitly understand that the embodiments described in the present application can be combined with other embodiments without conflict.

[0047] Unless otherwise defined, technical terms and scientific terms used in the present application shall have the ordinary meanings as understood by one of ordinary skill in the art to which this application pertains. The terms "a", "an", "one", "this", and similar referents in the context of describing the application are to be construed to be inclusive, not exclusive. The terms "comprise", "comprising", "comprises", "include", "including", "includes", "contain", "containing", "contains", and any variations thereof are intended to cover a non-exclusive inclusion, such that a process, method, system, product, or apparatus that comprises a list of steps or units uses "comprising" to indicate that it includes the listed steps or units, but is not limited to only those steps or units. The terms "connected", "coupled", and "pathway" are not limited to direct or physical connections, but can include indirect connections, unless otherwise explicitly stated. The term "plurality" means two or more. The term "and / or" describes associated objects in association relationships, which means that there are three relationships, for example, "A and / or B" can mean that A exists alone, A and B exist together, and B exists alone. The terms "first", "second", "third", and the like are only used to distinguish similar objects, and do not represent a specific order.

[0048] The method embodiment provided by the embodiment can be executed in a terminal, a computer or a similar computing device. Taking the running on the terminal as an example, Figure 1 is a hardware structure block diagram of the terminal of the high-speed rail line wind speed monitoring point layout optimization method of the embodiment. As shown in Figure 1 , the terminal can include one or more (only one in Figure 1 ) processor 102 (the processor 102 can include but is not limited to a processing device such as a microprocessor MCU or a programmable logic device FPGA) and a memory 104 for storing data. Optionally, the above terminal can also include a transmission device 106 for communication function and an input and output device 108. Those skilled in the art can understand that Figure 1 The structure shown is only schematic, which does not limit the structure of the above terminal. For example, the terminal can include more or less components than those shown in Figure 1 , or have a different configuration from Figure 1 .

[0049] The memory 104 can be used to store computer programs, such as software programs of application software and modules, for example, a computer program corresponding to the high-speed rail line wind speed monitoring point layout optimization method in the embodiments of the present application. The processor 102 executes various functions and data processing by running the computer program stored in the memory 104, that is, implements the above method. The memory 104 can include a high-speed random access memory, and can also include a non-volatile memory, such as one or more magnetic storage devices, flash memories, or other non-volatile solid-state memories. In some examples, the memory 104 can further include a memory remotely arranged with respect to the processor 102, and the remote memory can be connected to the terminal through a network. Examples of the above network include but are not limited to the Internet, an intranet, a local area network, a mobile communication network, and a combination thereof.

[0050] The transmission device 106 is used to receive or send data via a network. The specific examples of the above network can include a wireless network provided by a communication provider of the terminal. In one example, the transmission device 106 includes a network adapter (NIC), which can be connected to other network devices through a base station so as to communicate with the Internet. In one example, the transmission device 106 can be a radio frequency (RF) module, which is used to communicate with the Internet in a wireless manner.

[0051] The embodiment provides a high-speed rail line wind speed monitoring point layout optimization method, Figure 2 is a flowchart of the high-speed rail line wind speed monitoring point layout optimization method according to the embodiment of the present application, as shown in Figure 2 The flowchart includes the following steps:

[0052] In step S201, wind speed data of each monitoring point in an initial layout position is obtained; based on the wind speed data between each two monitoring points, a multi-dimensional correlation index set is generated;

[0053] The wind speed data of all monitoring points in the initial layout position along the high-speed rail is collected, for example, one year of historical wind speed data of each monitoring point is collected, one second is collected, and the original data amount is 1.23 billion. Then, the collected wind speed data is preprocessed to eliminate missing values (such as missing data in some time periods) and outliers (such as invalid records with wind speed ≤ 0 m / s), to ensure data integrity. The preprocessed wind speed data can also be aligned and reduced in dimension, for example, the data is smoothed by 10-minute average value, and 1.64 million valid data are formed after cleaning. The correlation between all monitoring points on the whole line is analyzed (such as 741 pairs formed by 39 monitoring points), and the Pearson correlation coefficient, Spearman rank correlation coefficient, Kendall rank correlation coefficient, cosine similarity, Euclidean distance and Canberra distance of the wind speed data sequence are calculated, to generate a multi-dimensional correlation index set containing the above six types of indexes.

[0054] The prior art only analyzes the similarity of adjacent monitoring points, and cannot identify small and medium-sized independent wind fields. The present application calculates the correlation of all monitoring point pairs, covers the whole line network correlation, reveals high-risk areas (such as valley outlets) and monitoring blind areas, and comprehensively evaluates the wind speed correlation between monitoring points through the multi-dimensional correlation index set, providing a rich data basis for subsequent analysis. The comprehensive use of multi-dimensional indexes can more accurately reflect the true situation of wind speed data, reduce the deviation that may be caused by a single index. Specifically, linear correlation (Pearson) quantifies the linear trend consistency of wind speed sequence, nonlinear correlation (Spearman, Kendall) is robust to non-normal distribution or outliers, cosine similarity measures the pattern direction consistency, and Euclidean distance quantifies the absolute difference. By analyzing the correlation between monitoring points, a scientific basis is provided for subsequent point optimization, which helps to improve the overall performance of the monitoring system.

[0055] In step S202, based on the multi-dimensional correlation index set, the weights of each index are allocated, the weight allocation result is generated, and the correlation weight between each pair of monitoring points is calculated; based on the weight allocation result and the correlation weight, an initial correlation network for the monitoring points is constructed;

[0056] Among them, the entropy weight method can be used to allocate weights to each index in the multi-dimensional correlation index set. The entropy weight method evaluates the value of each index by measuring its degree of differentiation. The higher the degree of differentiation, the greater the weight of the index. The specific steps include data standardization processing, calculation of index proportion, entropy value, and information utility value, and finally determining the weight of each index. Using the weight allocation result, the multi-dimensional correlation index between each pair of monitoring points is weighted and summed to obtain the correlation weight between each pair of monitoring points, that is, various correlation indexes are considered comprehensively to form the comprehensive correlation weight between each pair of monitoring points. Then, taking the wind speed monitoring point as the node and the correlation weight between the monitoring points as the edge, an initial correlation network with weight and without direction is constructed. This network reflects the correlation strength between the monitoring points based on wind speed data, providing a basis for subsequent network analysis and optimization.

[0057] This step allocates weights to each index, avoiding the influence of subjective judgment, making the weight allocation more objective and accurate. The weight is data-driven, avoiding artificial experience interference, and the weight allocation is dynamically adjusted with data (such as automatically increasing the distance measurement weight when the winter wind speed increases), which helps to more truly reflect the importance of each index in the correlation network; the correlation weight between the monitoring points is calculated based on the weighted sum, and the initial correlation network is constructed, so that the network can more accurately reflect the actual correlation between the monitoring points, providing a reliable basis for subsequent network correction and monitoring point layout optimization.

[0058] Step S203, among all connections of the initial correlation network, the connections with correlation weights lower than the preset weight threshold are identified as weak connections, and the weak connections are deleted to form a corrected target correlation network; based on the target correlation network, the initial layout position of the monitoring points is adjusted to the optimized layout position.

[0059] Among them, all connections in the initial correlation network are screened according to the preset weight threshold. The connections with correlation weights lower than the threshold are identified as weak connections. These weak connections may represent insignificant or redundant connections in the network, and have less impact on the overall network structure and function. Then, all edges identified as weak connections are deleted from the initial correlation network, thereby simplifying the network structure, removing redundant information, making the network more compact and efficient, and forming a corrected target correlation network: after deleting the weak connections, the remaining network structure is the corrected target correlation network, that is, this network focuses more on reflecting the significant and important correlation between the monitoring points. Finally, based on the corrected target correlation network, the initial layout position of the monitoring points is adjusted. This step can consider the centrality index (such as degree centrality, betweenness centrality, etc.) of the nodes in the network, evaluate the priority of the nodes, identify the weak links and key nodes in the network, and optimize the position of the monitoring points according to the evaluation result to enhance the coverage and monitoring ability of the network.

[0060] This step makes the network structure more compact by deleting weak connections, reduces redundant information, improves the overall efficiency of the network, helps to deliver information faster, and improves the response speed of the system; by removing redundant edges, the overall connectivity of the network is optimized, reducing the risk of network collapse caused by the failure of individual nodes or edges, enhancing the robustness and stability of the network, and improving the reliability of the system; based on the corrected target correlation network, the layout of the monitoring points is optimized, so that the monitoring points can more accurately capture extreme wind speed events, improving the spatial coverage accuracy and risk prediction efficiency of the strong wind warning system, providing stronger protection for high-speed rail operation safety; the corrected target correlation network provides clearer and more accurate network structure information for decision makers, which helps to develop a more scientific and reasonable monitoring point layout scheme, which provides strong support for the long-term planning and development of the wind speed monitoring system along the high-speed rail.

[0061] Through the above steps, compared with the traditional signal correlation method which only analyzes the linear similarity between adjacent monitoring points, it cannot identify small and medium scale independent wind fields, leading to monitoring blind area, and relying on a single Pearson coefficient, ignoring nonlinear relationships and subtle differences, the present application collects six types of indicators (Pearson, Spearman, Kendall, cosine similarity, Euclidean distance, Canberra distance) between each pair of monitoring points, fully quantifies the wind speed correlation from linear, nonlinear, pattern direction, and absolute difference dimensions, making up for the blind area of traditional methods in identifying complex terrain wind fields, and providing a multi-perspective data basis for subsequent network construction. Compared with the prior art which uses artificial experience weighting or relies only on a single indicator (such as cosine similarity), it is easily affected by subjective bias and cannot reflect the dynamic changes of the indicators, the present application dynamically allocates weights by calculating the information entropy of each indicator, ensuring the objectivity of data-driven, and the weighted undirected network constructed based on the weighted correlation weight upgrades the relationship between monitoring points from local linear correlation to global network correlation, breaking through the limitations of traditional "point-to-point" analysis. Compared with the traditional equidistant point layout which leads to coexistence of redundant alarm around the monitoring points and monitoring blind area at the exit of the valley, the present application sets a weight threshold, and the system identifies and deletes the weak connections in the initial correlation network, forming the corrected target correlation network. Based on this more compact and efficient network, the system optimizes and adjusts the initial layout position of the monitoring points, so that the monitoring points can more accurately capture extreme wind speed events, especially in high-risk areas, the monitoring density is increased, and the redundant monitoring in low-risk areas is reduced. This series of steps re-examines the monitoring point layout from a global perspective, effectively solving the problems of excessive alarm range, unreasonable resource allocation, and insufficient identification of high-risk areas caused by traditional local perspective layout, significantly improving the performance and efficiency of the wind speed monitoring system along the high-speed rail.

[0062] In some embodiments, based on the wind speed data between each pair of monitoring points, a multi-dimensional correlation index set is generated, including:

[0063] For the wind speed data between all monitoring points, the linear correlation coefficient, nonlinear rank correlation coefficient, similarity and distance metric indicators are calculated respectively, and based on the linear correlation coefficient, nonlinear rank correlation coefficient, similarity and distance metric indicators, a multi-dimensional correlation index set is generated.

[0064] Among them, the correlation analysis is carried out on all monitoring points in the whole line (such as 39 monitoring points forming 741 pairs). The index types include linear correlation (Pearson correlation coefficient), nonlinear correlation (Spearman rank coefficient, Kendall rank coefficient), similarity (cosine similarity) and distance metric (Euclidean distance, Canberra distance). The specific formula is as follows:

[0065] Pearson correlation coefficient (Pearson): Pearson correlation coefficient measures the linear correlation of two sets of wind speed data, and the value range is [-1, 1], the closer to 1 or -1, the stronger the linear correlation between the two monitoring points. The calculation of the correlation coefficient is shown in formula (1):

[0066]

[0067] In the formula, r(X, Y) is the correlation coefficient of data X and data Y, Var[X] is the variance of data X, and Var[Y] is the variance of data Y.

[0068] The calculation of the covariance value is shown in formula (2):

[0069]

[0070] In the formula, Cov(X, Y) is the covariance of data X and data Y, is the average value of data X, is the average value of data Y.

[0071] Spearman rank correlation coefficient (Spearman): Spearman rank coefficient is a non-parametric method, and the value range is [-1, 1], which measures the correlation between the ranks or orders of two variables. When the wind speed data is not completely linear or the distribution does not conform to the normal distribution assumption, Spearman coefficient can effectively measure the consistency of wind speed change trend between two places.

[0072]

[0073] In the formula, d i represents the difference between the ranks of the ith data pair, and n is the total number of observation samples.

[0074]

[0075] Kendall's rank correlation coefficient: Kendall's rank correlation coefficient is a nonparametric statistical method with a value range of [-1, 1], used to measure the correlation between the rankings of two sets of data. When wind speed data does not follow a normal distribution or contains outliers, Kendall's coefficient can assess the consistency of the wind speed intensity ranking between two locations and has good robustness to outliers. n pairs of wind speed observations (x1, y1), (x2, y2), ..., (x... n ,y n If the product (x) j -x i )×(y j -y i )>0 for If i, j = 1, 2, ..., n, then the pair (x i y i ) and (x j y j A set of pairs satisfies the cooperative property. Conversely, if they do not, they are said to be uncooperative. All data may share a common set of pairs. Yes, use N c For the number of logarithms in the same direction, N c It is a reverse logarithm.

[0076] In the formula, S = N c -N d -1 ≤ τ ≤ 1. If all pairs of numbers are consistent, then N c = n(n-1) / 2, N d =0, τ=1 indicates that the two sets of data are positively correlated; if all pairs are opposite, then N c =0, N d =n(n-1) / 2, τ=-1, indicating that the two sets of data are negatively correlated; when τ is zero, it indicates that the two sets of wind monitoring data are not correlated.

[0077] Cosine similarity: Cosine similarity measures the similarity between two sequences by calculating the cosine of the angle between them in n-dimensional space. The value ranges from [0,1]. Cosine similarity can measure the directional consistency and pattern similarity between two wind speed sequences. It is not sensitive to the size of the data and is suitable for comparing the overall consistency of wind speed patterns.

[0078]

[0079] Euclidean Distance: Euclidean distance measures the straight-line distance between two points in multi-dimensional space, reflecting the absolute distance between them. In wind speed correlation analysis, Euclidean distance can be used to quantify the absolute difference between the wind speed time series of two places. Euclidean distance or Euclidean metric is the straight-line distance between two points in Euclidean space. Using this distance, Euclidean space becomes a metric space, and the associated norm is called the Euclidean norm.

[0080]

[0081] Canberra Distance: Canberra distance assumes that variables are independent of each other and does not consider the correlation between variables. It is not sensitive to the dimension of data and is sensitive to small differences. When you want to analyze the small changes in wind speed differences in detail, Canberra distance is more suitable.

[0082]

[0083] Finally, a multi-dimensional correlation matrix containing the above six types of indicators is generated (as shown in the following table), that is, a multi-dimensional correlation indicator set.

[0084]

[0085] This embodiment can comprehensively evaluate the wind speed correlation between monitoring points by calculating multiple correlation indicators, including linear relationship, nonlinear relationship, similarity and distance difference, etc. The Pearson correlation coefficient measures the strength and direction of the linear relationship between two variables. The Spearman rank correlation coefficient and Kendall rank correlation coefficient are used to compare the ranks (i.e. order positions) of data points to evaluate nonlinear correlation. The cosine similarity is used to measure the direction consistency and pattern similarity between wind speed data sequences of each pair of monitoring points. The Euclidean distance and Canberra distance are used to quantify the absolute difference and relative difference between wind speed data of each pair of monitoring points. The multi-dimensional correlation indicator set containing the above indicators provides more rich information for monitoring point layout optimization, which helps to more accurately identify high-risk areas and weak links, thereby improving the pertinence and effectiveness of monitoring point layout.

[0086] In some embodiments, based on the multi-dimensional correlation indicator set, the weights of each indicator are assigned, and a weight assignment result is generated, including:

[0087] The indicators in the multi-dimensional correlation indicator set are standardized, the information entropy and information utility value corresponding to each indicator are calculated, and based on the information utility value, the weight assignment result of each indicator is determined.

[0088] Wherein, each index in the multi-dimension correlation index set is standardized to eliminate the dimensional difference between different indexes. The standardization is usually achieved by subtracting the mean value of each index from its value and dividing by the standard deviation, so that all indexes have the same scale. For each standardized index, its information entropy is calculated. Information entropy is an index that measures the uncertainty of data, which reflects the richness of information in the index. The greater the information entropy, the richer the information contained in the index, and the greater the contribution to the overall result. Based on the information entropy, the information utility value of each index is calculated. The information utility value is the complement of the information entropy, which is used to measure the effectiveness of the index in the overall evaluation. The higher the information utility value, the greater the importance of the index in the evaluation. According to the information utility value of each index, the weight allocation result of each index is determined. The weight allocation result is usually the normalization of the information utility value of all indexes, so that the sum of all weights is 1, and the specific weight value of each index is obtained, that is, the weight allocation result of each index is determined.

[0089] The embodiment determines the weight by calculating the information entropy and the information utility value, avoids the influence of subjective judgment, and makes the weight allocation more objective and accurate. The weight allocation result is completely based on the data itself, reflects the actual information distribution in the data, makes the subsequent analysis and optimization more in line with the actual situation, and the reasonable weight allocation can more accurately reflect the importance of each index in the overall evaluation, thereby improving the accuracy of subsequent analysis and optimization. By optimizing the weight allocation, the monitoring data can be more effectively utilized, the performance and efficiency of the high-speed rail along the wind speed monitoring system can be improved, and the overall stability and safety of the system can be enhanced.

[0090] In some embodiments, the preset weight threshold is determined according to the obtained network structure index, and the network structure index includes the clustering coefficient, network efficiency, average degree, and number of connected components of the initial association network.

[0091] Among them, four key network structure indicators are collected for the initial correlation network: clustering coefficient, network efficiency, average degree, and connected components. When deleting weak connections in the initial correlation network, the changes in these network structure indicators are recorded and analyzed step by step. Specifically, according to the order of correlation weight from low to high, the edges are deleted one by one, and the changes in clustering coefficient, network efficiency, average degree, and connected components after each edge deletion are observed and recorded. The weight threshold is determined according to the mutation point or critical value of the network structure indicators. For example, when the clustering coefficient or network efficiency starts to decrease significantly, or the number of connected components increases sharply, it can be considered that the current deleted edge has an important impact on the network structure, and the correlation weight value at this time can be used as the preset weight threshold. By calculating the clustering coefficient to reflect local closeness, network efficiency to reflect global information transmission ability, average degree and connected components to ensure network connectivity, and by determining the threshold based on the mutation points of clustering coefficient and network efficiency, both efficiency and connectivity are considered. The specific formula is as follows:

[0092] Clustering Coefficient: Clustering coefficient is an indicator that measures the tightness of connections between the neighbors of a node. A network with high clustering coefficient indicates that nodes tend to form tightly connected groups. In a correlation network, a high clustering coefficient means that some nodes in the wind speed monitoring network have high correlation in their wind speed sequences. For node v, the formula for calculating the clustering coefficient C(v) is:

[0093]

[0094] where E v is the number of edges actually existing between the neighbors of node v, and D v is the degree of node v. This formula measures the tightness of connections between the neighbors of a node.

[0095] Network Efficiency: Global network efficiency measures the efficiency of information flow on the average shortest path in the network. It evaluates the communication efficiency and transmission speed of the entire network. A high-efficiency network means faster information transmission between nodes and stronger correlation between nodes. The formula for calculating global network efficiency E glob is:

[0096]

[0097] where d(i, j) is the shortest path length between nodes i and j, and N is the total number of nodes in the network. This measure reflects the overall information transmission efficiency of the network.

[0098] Average Degree: Average degree is a basic indicator in network analysis, which measures the average number of connections of each node in the network, reflecting the overall connectivity of the network. A higher average degree usually indicates that the nodes in the network are more associated with each other, and the network as a whole has higher connectivity and information transmission capacity. Average degree <k>The calculation formula is:

[0099]

[0100] In the formula, <k>where M is the total number of edges in the graph, and N is the total number of nodes in the graph.

[0101] Connected Components: Connected components refer to the largest subgraph in which all nodes are connected to each other, and this connected component does not contain any other connected nodes. A connected undirected graph has only one connected component, i.e., a connected subgraph containing all nodes. The number of connected components can be used to measure the overall connectivity of a network graph. If there are isolated nodes or subgraphs in the graph, the number of connected components of the graph will increase.

[0102] Connected components are an important indicator in network analysis, which can reflect the connectivity and structural characteristics of the network. By analyzing the number of connected components, the robustness and stability of the network can be evaluated. If the number of connected components increases significantly after deleting a few nodes or edges, it indicates that the network is relatively fragile.

[0103] The embodiment determines the weight threshold based on network structure indicators, making the setting of the threshold more scientific and reasonable, avoiding subjective speculation, and improving the accuracy and reliability of the weight threshold. Different network structures have different characteristics, and the weight threshold is determined by dynamically analyzing the changes in network structure indicators, making the method adaptable to networks with different characteristics and improving the universality and adaptability of the method. A reasonable weight threshold can effectively delete redundant edges and retain key edges that have a greater impact on network structure and function, thereby forming a more compact and efficient modified target association network, which helps to improve the effect of subsequent monitoring point layout optimization and enables the monitoring point to capture extreme wind speed events more accurately.

[0104] In some embodiments, based on the target association network, the initial layout position of the monitoring point is adjusted to the optimized layout position, including:

[0105] For each monitoring point of the target association network, a monitoring point priority ranking result is calculated; and based on the target association network and the monitoring point priority ranking result, the initial layout position of the monitoring point is adjusted to the optimized layout position.

[0106] The centrality indexes of each monitoring point in the target correlation network are calculated, such as degree centrality, betweenness centrality, closeness centrality, etc. According to the calculated centrality indexes, a comprehensive score is given to each monitoring point, and the higher the score, the higher the priority. Then, the monitoring points are prioritized according to the scores. According to the priority ranking result, the key area in the network is identified, i.e. the area containing high-priority monitoring points. The number of monitoring points in the key area is increased or the position of the existing monitoring points is adjusted to improve the capture ability of extreme wind speed events. In the low-priority area, the number of monitoring points is considered to be reduced or their positions are adjusted to avoid waste of resources. In addition, when adjusting the layout of the monitoring points, the connectivity of the target correlation network is ensured to be unaffected, so that information can be effectively transmitted in the network. According to the above optimization layout strategy, the initial layout position of the monitoring points is adjusted to the optimized layout position.

[0107] According to the priority ranking result of the monitoring points, the monitoring resources are reasonably allocated, the over-investment in low-risk areas is avoided, the resource utilization efficiency is improved, the clear priority ranking of the monitoring points and the optimized layout position provide clear data support for decision makers, which helps to develop more scientific and reasonable monitoring and early warning strategies; by prioritizing the layout of high-priority monitoring points, extreme wind speed events can be captured more quickly, the response speed and efficiency of the monitoring system are improved, and the layout of the monitoring points can be dynamically adjusted according to different network structures and environmental characteristics to adapt to the complex and variable wind speed monitoring requirements along the high-speed rail.

[0108] In some embodiments, the priority ranking result of the monitoring points of the target correlation network is calculated, including:

[0109] The multi-dimensional node importance evaluation indexes of the monitoring points of the target correlation network are calculated;

[0110] The actual values of the node indexes are calculated based on the multi-dimensional node importance evaluation indexes;

[0111] The priority ranking result of the monitoring points is determined by calculating the total deviation value of the actual values of the node indexes and the preset ideal values of the indexes.

[0112] Among them, the multi-dimensional node importance evaluation index includes degree centrality, betweenness centrality, closeness centrality, eigenvector centrality, PageRank, and K-shell centrality. By using the six indexes of degree centrality, betweenness centrality, closeness centrality, eigenvector centrality, PageRank centrality and K-shell centrality, the importance and function of the monitoring point in the whole system can be comprehensively evaluated. Degree centrality and eigenvector centrality reveal the number of direct connections and connection quality of the monitoring point, highlighting its activity and influence in the network. Betweenness centrality and closeness centrality reflect the control and access efficiency of the monitoring point in the information flow, which is crucial for rapid response and data transmission. PageRank centrality emphasizes the trustworthiness and influence of the wind speed monitoring point along the high-speed rail, while K-shell centrality identifies the stable nodes in the network core. Using these indexes to determine the priority of the newly added monitoring points can accurately identify the weak links in the network, and then effectively optimize the overall layout of the monitoring points and improve the efficiency of strong wind monitoring. The specific formulas are as follows:

[0113] Degree Centrality: Degree centrality reflects how many other nodes a node is directly connected to. Nodes with high degree centrality are usually considered "hot spots" or key active nodes in the network. For node v, the calculation formula of degree centrality DC is:

[0114]

[0115] In the formula, d(v) is the degree of node v, that is, the number of edges connected to node v, and N is the total number of nodes in the network.

[0116] Betweenness Centrality: Betweenness centrality shows the importance of a node in connecting different parts of the network. Nodes with high betweenness centrality control the connection between different nodes or groups of nodes in the network, and have stronger control and influence on the whole network. For node v, the calculation formula of betweenness centrality BC is:

[0117]

[0118] In the formula, σ st is the number of shortest paths from node s to node t, and σ st (v) is the number of shortest paths from node s to node t passing through node v. This index measures the frequency of node v appearing in the shortest paths between all node pairs.

[0119] Closeness Centrality: Closeness Centrality measures the average distance of a node to all other nodes in the network, with higher closeness centrality indicating faster access to other nodes in the network, which means they have an advantage in quickly spreading information or resources. For a node v, the formula for calculating closeness centrality is:

[0120]

[0121] where d(v, u) is the shortest path length between nodes v and u, and N is the total number of nodes in the network. This measure reflects the average distance of v to all other nodes in the network.

[0122] Eigenvector Centrality: Eigenvector Centrality reflects the influence of a node, with the importance of a node not only depending on the number of its direct connections, but also on the importance of these neighbor nodes in the global network. If a node is connected to more core nodes with high influence, it indicates that it occupies a more important strategic position in the global structure of the network. For node i, the formula for calculating eigenvector centrality EC(i) is:

[0123]

[0124] where a ij is the connection weight between nodes i and j (in the adjacency matrix), and λ is the eigenvalue corresponding to node i in the adjacency matrix. The importance of a node depends on the importance of the nodes it is connected to.

[0125] Page Rank Centrality (PR): Page Rank Centrality is a variant of eigenvector centrality designed to handle the connection structure in the network, considering the directionality and weight of connections, and is particularly effective in assessing the importance of nodes, helping to identify the most valuable or most trusted nodes in the network. The formula for calculating Page Rank is:

[0126]

[0127] where PR(p i ) is the Page Rank value of node p i , M(p i ) is the set of all nodes pointing to node p i , L(p j ) is the number of outgoing connections of node p j , d is the damping factor, usually set to 0.85, and N is the total number of pages in the network.

[0128] K-shell Centrality: Based on the hierarchical position of nodes in the network, the higher the core level of the node in the network, the larger the k-shell value. It can be used to identify the core and edge structure of the network, and then understand the stability and resistance of the network. The k-shell decomposition method does not have a concise formula, it is based on iterative stripping of nodes with degree less than k in the network. Each node is assigned to a k-shell, and this value k is the maximum value that the node can be stripped.

[0129] For each monitoring point, the actual value of the above multi-dimensional node importance evaluation index is calculated respectively. For example, the degree centrality can be directly counted, the betweenness centrality needs to be calculated by path algorithm, the eigenvector centrality needs to be iteratively solved, the closeness centrality needs to be calculated, and the clustering coefficient needs to be calculated. According to the monitoring requirements and network characteristics, set the ideal value for each index, for example, a high-risk area may require a higher betweenness centrality to control information propagation, and a low-risk area can accept a lower closeness centrality. For each monitoring point, the deviation of the actual value of each index from the ideal value is calculated, which can be absolute deviation, relative deviation or weighted deviation. The total deviation value of each monitoring point is sorted, and the smaller the deviation value indicates that the monitoring point is closer to the ideal state, and the higher the priority, for example, the total deviation value is obtained by using the weighted summation method to integrate the deviation of each index, and then sorted to determine the priority ranking result of the monitoring point.

[0130] The embodiment comprehensively evaluates the importance of the monitoring point in the network from multiple angles such as local to global, direct to indirect, and propagation speed to structural characteristics through multi-dimensional node importance evaluation indexes, ensuring that the evaluation result is comprehensive and accurate; based on the deviation between the actual value and the ideal value calculated by the multi-dimensional index, the one-sidedness of single index evaluation is avoided, the scientificity and rationality of the priority ranking are improved, the evaluation index and the ideal value can be dynamically adjusted according to different network structures and monitoring requirements, and the complex and variable high-speed rail along the wind speed monitoring network environment is adapted; through the clear priority ranking result, a clear basis is provided for monitoring and early warning, which helps to concentrate limited resources on high-priority monitoring points, improves monitoring efficiency and early warning accuracy; through comprehensive evaluation of node importance and reasonable sorting, it is helpful to quickly identify and adjust other high-priority monitoring points when some monitoring points fail, and enhance the robustness and stability of the system.

[0131] In some embodiments, based on the multi-dimensional node importance evaluation index, the actual value of the node index is calculated, including:

[0132] Based on the wind speed data, the coefficient of variation is calculated; based on the multi-dimensional correlation index set and the coefficient of variation, the importance index weight of the multi-dimensional node importance evaluation index is determined;

[0133] Based on the multi-dimensional node importance evaluation index and the importance index weight, the actual value of the node index is calculated.

[0134] wherein, for each monitoring point's wind speed data, its mean (μ) and standard deviation (σ) are calculated. The mean reflects the average level of the monitoring point's wind speed, while the standard deviation measures the dispersion of the wind speed data. Then the coefficient of variation is calculated, which is the ratio of the standard deviation to the mean, i.e. CV = σ / μ. The larger the coefficient of variation, the greater the volatility of the monitoring point's wind speed data, and the more likely it has a higher monitoring value. The multi-dimensional correlation index set includes linear correlation coefficient, nonlinear rank correlation coefficient, similarity and distance metric indicators, etc., which are used to evaluate the correlation between monitoring points. The coefficient of variation is used as one dimension to evaluate the importance of monitoring points, combined with other correlation indicators. The monitoring point with a larger coefficient of variation may play a more critical role in the network, as its wind speed changes may have a greater impact on other monitoring points. The entropy weight method or other objective weighting methods are used to assign weights to each multi-dimensional node importance evaluation indicator, combined with the multi-dimensional correlation index set and the coefficient of variation. The weight assignment should reflect the relative contribution of each indicator in evaluating the importance of the node, ensuring the objectivity and accuracy of the evaluation results. The actual value of each monitoring point's multi-dimensional node importance evaluation indicator is multiplied by its corresponding importance indicator weight, and then the weighted sum is calculated to obtain the node indicator actual value of each monitoring point.

[0135] This step can more comprehensively evaluate the importance of monitoring points in the network by considering the coefficient of variation of wind speed data and the multi-dimensional correlation index set, improving the accuracy of the evaluation results; since the weight assignment is dynamically calculated based on actual data, this method can adapt to changes in network structure, ensuring reasonable evaluation results under different circumstances; the explicit node indicator actual value provides strong support for the optimization of monitoring point layout. Decision makers can adjust the layout of monitoring points based on the evaluation results to improve the overall performance and efficiency of the monitoring system; by focusing on high importance monitoring points first, limited resources can be concentrated in key areas, improving resource utilization efficiency and reducing monitoring costs; by considering multiple evaluation indicators and weight assignment methods, the robustness and stability of the monitoring system can be enhanced, ensuring high monitoring performance even when some monitoring points fail.

[0136] In some embodiments, based on the target correlation network and the monitoring point priority ranking result, the initial layout position of the monitoring point is adjusted to the optimized layout position, including:

[0137] The target correlation network is divided into modules to obtain a correlation network module, and the correlation network module is matched with the terrain features based on the obtained terrain features of the monitoring points to obtain a terrain feature verification result;

[0138] Based on the target correlation network, the monitoring point priority ranking result and the terrain feature verification result, the initial layout position of the monitoring point is adjusted to the optimized layout position.

[0139] Wherein, the community discovery algorithm (such as Louvain algorithm, Girvan-Newman algorithm, etc.) is used to divide the target correlation network into modules, and the network is divided into multiple correlation network modules, which reflect the wind speed correlation characteristics of different regions in the network. Each module represents a group of monitoring points with similar wind speed characteristics and similar geographical location, which shows high correlation in the network. Collect the terrain feature data of each monitoring point, including altitude, terrain undulation, surrounding obstacles (such as mountains, buildings) and other information. Match each correlation network module with the corresponding monitoring point terrain features, analyze the influence of terrain factors on wind speed correlation, for example, some modules may correspond to different terrains such as valleys, plains or mountainous areas, and these terrain factors will affect the distribution and change of wind speed. Through comparative analysis, the consistency of correlation network module division and terrain features is verified to ensure the rationality of module division, and the terrain feature verification result is obtained. Combined with the structure characteristics of the target correlation network, the priority ranking result of the monitoring point and the terrain feature verification result, the layout of the monitoring point is comprehensively evaluated. According to the evaluation result, the initial layout position of the monitoring point is adjusted. Specifically, monitoring points can be added or adjusted near key modules and high-priority monitoring points to improve the coverage and accuracy of the monitoring system. At the same time, considering the influence of terrain factors on wind speed monitoring, it is necessary to avoid excessive deployment of monitoring points in areas with unclear wind speed changes. After layout adjustment, module division, priority ranking and terrain feature verification can be performed again to iteratively optimize the layout of monitoring points until a satisfactory monitoring effect is achieved.

[0140] The embodiment can more accurately identify key areas of wind speed change by module division and terrain feature matching, thereby increasing monitoring points in these areas and improving the accuracy of the monitoring system. According to the priority ranking result of the monitoring point and the terrain feature verification result, monitoring resources can be reasonably allocated to avoid excessive investment in low-risk areas and improve resource utilization efficiency. The method takes into account the influence of terrain factors on wind speed monitoring, so that the monitoring system can better adapt to wind speed changes under different terrain conditions, improving the adaptability and robustness of the system. Through comprehensive evaluation and iterative optimization, a scientific basis is provided for the layout adjustment of monitoring points, which helps decision-makers make more reasonable decisions and improves the overall performance of the monitoring system. By optimizing the layout of monitoring points, unnecessary monitoring equipment investment can be reduced, monitoring costs can be reduced, and monitoring efficiency can be improved, which helps to achieve sustainable development of high-speed rail wind speed monitoring.

[0141] The preferred embodiments of the present application are described and explained below.

[0142] The preferred embodiment takes the historical wind speed data of each wind monitoring point arranged along the railway as input, and outputs the wind monitoring point priority ranking, which will be used to guide the deployment optimization of the monitoring point. After preprocessing the input historical data, the method uses multiple measurement methods (linear correlation, nonlinear correlation, similarity and distance measurement) to test the correlation, and integrates the correlation indicators through entropy weighting method to build the correlation edges between monitoring points, and then performs sensitivity analysis to build an effective correlation network. The nodes (wind speed monitoring points) in the correlation network are taken as evaluation objects, the centrality indicators of the nodes in the network are calculated, and the CRITIC method is applied to weight the indicators, combined with the MARICA method to evaluate the priority ranking of the nodes, and identify the weak links of the network to provide basis for the site selection of new monitoring points. Figure 3 is a preferred flowchart of the high-speed railway along the wind speed monitoring point layout optimization method according to the embodiments of the present application, as Figure 3 shown, the high-speed railway along the wind speed monitoring point layout optimization method comprises the following steps:

[0143] Step S301, data preprocessing. Each wind speed monitoring point is regarded as a node in the network, and one year of data is collected as the basic input. The initial data of each monitoring point is sampled every ten minutes, and the missing data is deleted and aligned.

[0144] Step S302, calculate the correlation between monitoring points. The wind speed data of all monitoring points on the whole line is tested for correlation, and the Pearson correlation coefficient, Spearman rank correlation coefficient, Kendall rank correlation coefficient, cosine similarity, Euclidean distance and Canberra distance of the wind speed sequence are calculated.

[0145] Step S303, build the correlation network. The wind speed monitoring points are taken as nodes, and the correlation strength between the monitoring points is taken as edges. The 6 types of correlation indicators of each pair of monitoring points in step 2 are integrated by entropy weighting method, the correlation weight between the monitoring points is determined, and a weighted undirected original correlation network is established.

[0146] Step S304, network correction. The edges in the network are deleted in order from low to high according to the correlation weight, the dynamic changes of the 4 structural indicators of the network clustering coefficient, network efficiency, average degree and connected component number are observed and analyzed, the appropriate weight threshold is determined according to the critical transition point of the network structure, the redundant edges incorrectly constructed are removed according to the weight threshold, and the corrected correlation network is formed.

[0147] Step S305, verify the effectiveness of the constructed correlation network through network modularity analysis. This process divides the nodes in the network into communities, and combines the topographic conditions of the monitoring point layout to evaluate the rationality and effectiveness of the network structure.

[0148] Step S306, evaluating node priority. The nodes in the associated network are taken as evaluation objects, and the CRITIC method is applied to weight the six centrality indexes of the nodes, including degree centrality, betweenness centrality, closeness centrality, eigenvector centrality, Page Rank centrality and K-shell centrality. Then, the MARICA multi-attribute decision method is used to calculate the total deviation of the actual state and the ideal state of each node, and further evaluate the priority of each node.

[0149] Step S307, according to the priority ranking of the nodes, identifying the weak links in the network, and formulating an optimized monitoring point layout scheme to enhance the coverage and monitoring capability of the network.

[0150] In the above step S302, the calculation process of the correlation index is as follows:

[0151] The correlation index is used to measure the similarity of the wind speed sequences recorded by two monitoring points in terms of long-term trend, etc. Considering the multi-scale time characteristics, randomness, spatial heterogeneity and suddenness of the wind speed along the high-speed rail, multiple correlation relationships are mined from the perspectives of linear correlation, nonlinear correlation, similarity and distance measurement, and a correlation connection with mixed properties is constructed. The correlation connection is used to determine the priority of the new monitoring point.

[0152] Pearson correlation coefficient (Pearson): The Pearson correlation coefficient measures the linear correlation degree of two sets of wind speed data, and the value range is [-1, 1]. The closer the value is to 1 or -1, the stronger the linear correlation between the two monitoring points. The calculation of the correlation coefficient is shown in formula (1):

[0153]

[0154] In the formula, r(X, Y) is the correlation coefficient of data X and data Y, Var[X] is the variance of data X, and Var[Y] is the variance of data Y.

[0155] The calculation of the covariance value is shown in formula (2):

[0156]

[0157] In the formula, Cov(X, Y) is the covariance of data X and data Y, is the average value of data X, is the average value of data Y.

[0158] Spearman rank correlation coefficient: The Spearman rank coefficient is a non-parametric method with a value range of [-1, 1]. It measures the correlation between the ranks or order of two variables. When wind speed data are not completely linear or the distribution does not conform to the normal distribution assumption, the Spearman coefficient can effectively measure the consistency of wind speed change trends between two locations.

[0159]

[0160] In the formula, d i This represents the difference in rank between the i-th data pairs, where n is the total number of observed samples.

[0161]

[0162] Kendall's rank correlation coefficient: Kendall's rank correlation coefficient is a nonparametric statistical method with a value range of [-1, 1], used to measure the correlation between the rankings of two sets of data. When wind speed data does not follow a normal distribution or contains outliers, Kendall's coefficient can assess the consistency of the wind speed intensity ranking between two locations and has good robustness to outliers. n pairs of wind speed observations (x1, y1), (x2, y2), ..., (x... n ,y n If the product (x) j -x i )×(y j -y i )>0 for If i, j = 1, 2, ..., n, then the pair (x i y i ) and (x j y j A set of pairs satisfies the cooperative property. Conversely, if they do not, they are said to be uncooperative. All data may share a common set of pairs. Yes, use N c For the number of logarithms in the same direction, N c It is a reverse logarithm.

[0163] In the formula, S = N c -N d -1 ≤ τ ≤ 1. If all pairs of numbers are consistent, then N c = n(n-1) / 2, N d =0, τ=1 indicates that the two sets of data are positively correlated; if all pairs are opposite, then N c =0, N d =n(n-1) / 2, τ=-1, indicating that the two sets of data are negatively correlated; when τ is zero, it indicates that the two sets of wind monitoring data are not correlated.

[0164] Cosine Similarity: Cosine similarity measures the similarity between two sequences by calculating the cosine of the angle between them in n-dimensional space, with a value range of [0, 1]. Cosine similarity can measure the directional consistency and pattern similarity of two wind speed sequences, and is not sensitive to data size, making it suitable for comparing the overall consistency of wind speed patterns.

[0165]

[0166] Euclidean Distance: Euclidean distance measures the straight-line distance between two points in multi-dimensional space, reflecting their absolute distance. In wind speed correlation analysis, Euclidean distance can be used to quantify the absolute difference between wind speed time series at two locations. Euclidean distance or Euclidean metric is the straight-line distance between two points in Euclidean space. Using this distance, Euclidean space becomes a metric space, and the associated norm is called the Euclidean norm.

[0167]

[0168] Canberra Distance: Canberra distance assumes independence between variables and does not consider the correlation between variables. It is not sensitive to the magnitude of data and is sensitive to small differences. When detailed analysis of small changes in wind speed differences is required, Canberra distance is more suitable.

[0169]

[0170] In the above step S304, the calculation process of the network structural index is as follows:

[0171] The network structural index can provide quantitative basis for the structural evolution analysis of real networks. During the process of deleting edges, by monitoring the dynamic evolution characteristics of the network structural index, the overall response mode of the topological structure can be quantitatively evaluated, and the optimal deletion threshold of redundant edges can be determined.

[0172] Clustering Coefficient: Clustering coefficient is an index that measures the tightness of connections between the neighbors of a node. A network with high clustering coefficient indicates that nodes tend to form tightly connected groups. In a correlation network, a high clustering coefficient means that the wind speed sequences of certain nodes in the wind speed monitoring network have high correlation. For a node v, the formula for calculating the clustering coefficient C(v) is:

[0173]

[0174] In the formula, E v is the number of edges actually existing between the neighbors of node v, and D v is the degree of node v. This formula measures the tightness of the connections between the neighbors of a node.

[0175] Network Efficiency: Global network efficiency measures the efficiency of information flow in the network along the average shortest paths. Evaluating the communication efficiency and transmission speed of the entire network, a high-efficiency network means faster information transfer between nodes and stronger correlation between nodes. The formula for calculating the global network efficiency E glob is:

[0176]

[0177] where d(i, j) is the shortest path length between nodes i and j, and N is the total number of nodes in the network. This measure reflects the overall information transmission efficiency of the network.

[0178] Average Degree: Average degree is a basic indicator in network analysis, used to measure the average number of connections of each node in the network, reflecting the overall connectivity of the network. A higher average degree usually indicates that nodes in the network are more associated with each other, and the network as a whole has higher connectivity and information propagation ability. The average degree <k>The calculation formula is:

[0179]

[0180] In the formula, <k>where M is the total number of edges in the graph, and N is the total number of nodes in the graph.

[0181] Connected Components: A connected component refers to the largest subgraph in which all nodes are connected to each other, and this connected component does not contain any other connected nodes. A connected undirected graph has only one connected component, i.e., a connected subgraph containing all nodes. The number of connected components can be used to measure the overall connectivity of a network graph, and if there are isolated nodes or subgraphs in the graph, the number of connected components of the graph will increase.

[0182] Connected components are an important indicator in network analysis, which can reflect the connectivity and structural characteristics of the network. By analyzing the number of connected components, the robustness and stability of the network can be evaluated. If the number of connected components increases significantly after deleting a few nodes or edges, it indicates that the network is relatively fragile.

[0183] In the above step S306, the calculation process of the node centrality indicator is as follows:

[0184] In the strong wind monitoring system, by using degree centrality, betweenness centrality, closeness centrality, eigenvector centrality, Page Rank centrality and K-shell centrality, the importance and function of the monitoring points in the entire system can be comprehensively evaluated. Degree centrality and eigenvector centrality reveal the number of direct connections and connection quality of the monitoring points, highlighting their activity and influence in the network. Betweenness centrality and closeness centrality reflect the control power and access efficiency of the monitoring points in the information flow, which are crucial for rapid response and data transmission. Page Rank centrality emphasizes the trustworthiness and influence of the wind speed monitoring points along the high-speed rail, while K-shell centrality identifies the stable nodes in the network core. Using these indicators to determine the priority of new monitoring points can accurately identify the weak links in the network, and effectively optimize the overall layout of the monitoring points, improving the efficiency of strong wind monitoring.

[0185] Degree Centrality: Degree centrality reflects how many other nodes a node is directly connected to. Nodes with high centrality are usually considered "hot spots" or key active nodes in the network. For a node v, the calculation formula of degree centrality DC is:

[0186]

[0187] where d(v) is the degree of node v, i.e., the number of edges connected to node v, and N is the total number of nodes in the network.

[0188] Betweenness Centrality: Betweenness Centrality shows the importance of a node in connecting different parts of the network. Nodes with high betweenness centrality control the connections between different nodes or groups of nodes, and have more control and influence over the network as a whole. The formula for calculating the betweenness centrality BC for a node v is:

[0189]

[0190] where σ st is the number of shortest paths from node s to node t, σ st (v) is the number of shortest paths from node s to node t that pass through node v. This measure reflects the frequency of occurrence of node v in the shortest paths between all pairs of nodes.

[0191] Closeness Centrality: Closeness Centrality measures the average distance of a node to all other nodes in the network. Nodes with high closeness centrality can access other nodes in the network more quickly, which means they have an advantage in quickly spreading information or resources. The formula for calculating the closeness centrality for a node v is:

[0192]

[0193] where d(v, u) is the shortest path length between nodes v and u, and N is the total number of nodes in the network. This measure reflects the average distance of v to all other nodes in the network.

[0194] Eigenvector Centrality: Eigenvector Centrality reflects the influence of a node, and the importance of a node not only depends on the number of its direct connections, but also on the importance of these neighbor nodes in the global network. If a node is connected to more core nodes with high influence, it indicates that it occupies a more important strategic position in the global structure of the network. The formula for calculating the eigenvector centrality EC(i) for a node i is:

[0195]

[0196] where a ij is the connection weight between nodes i and j (in the adjacency matrix), and λ is the eigenvalue corresponding to node i in the adjacency matrix. The importance of a node depends on the importance of the nodes it is connected to.

[0197] Page Rank Centrality (PR): Page Rank Centrality is a variant of Eigenvector Centrality designed to handle the connectivity structure in networks, considering the directionality and weight of connections, and is particularly effective in assessing the importance of nodes, helping to identify the most valuable or most trusted nodes in the network. The calculation formula of Page Rank is:

[0198]

[0199] where PR(p i ) is the Page Rank value of node p i , M(p i ) is the set of all nodes pointing to node p i , L(p j ) is the number of outgoing connections of node p j , d is the damping factor, usually set to 0.85, and N is the total number of pages in the network.

[0200] K-shell Centrality: According to the hierarchical position of nodes in the network, the higher the core level of nodes in the network, the larger the k-shell value. It can be used to identify the core and edge structure of the network, and further understand the stability and resistance of the network. The k-shell decomposition method does not have a concise formula, it is based on iterative stripping of nodes with degree less than k in the network. Each node is assigned to a k-shell, and this value k is the maximum value that the node can be stripped.

[0201] In the above step S303, the application process of the entropy weight method is as follows:

[0202] The entropy weight method evaluates the value of the index by measuring the degree of differentiation. The higher the dispersion of the measurement value, the higher the differentiation of the index, and the more information that can be derived, so the index should be given a higher weight, and vice versa.

[0203] When facing different dimensional wind speed correlation indicators, the entropy weight method ensures that different dimensional correlation test values can be reasonably compared and comprehensively judged through standardization processing, and determines the size of the weight according to the difference between the indicators, with strong stability. The main steps of the entropy weight method are as follows:

[0204] S303-1, initial data standardization processing to eliminate the dimensional differences between different indicators:

[0205]

[0206] where X ij is the value of the jth sample on the ith index, and are the minimum and maximum values of the jth index, respectively.

[0207] S303-2, calculate the proportion P of the index ij :

[0208]

[0209] where P ij is the proportion of the ith sample on the jth index.

[0210] S303-3, calculate the entropy value E j :

[0211]

[0212] where E j is the information entropy of the jth index, and the constant n is the number of samples.

[0213] S303-4, calculate the information utility value (anti-entropy value) d of each index j :

[0214] d j = 1-E j #(19)

[0215] S303-5 calculate the jth index weight ω j :

[0216]

[0217] where w j is the weight of the jth index, and m is the number of indexes.

[0218] In the above step S306, the application process of CRITIC method is as follows:

[0219] CRITIC (Criteria Importance Through Intercriteria Correlation) method is to allocate index weight according to the information of each index and the correlation between each index. The wind speed data of each monitoring point in the high-speed railway strong wind monitoring system is related to each other, and the periodic variation law of wind speed brings correlation to the index. Based on the above two points, the improved parameter coefficient CRITIC method is selected to determine the weight of each evaluation index, and the correlation information between different indexes and the variability coefficient v j is an important factor for calculating the comprehensive information amount c j . The main steps are as follows:

[0220] S306-11, calculate the standard deviation σ of each evaluation index j :

[0221]

[0222] wherein x ij is the value of the i-th sample on the j-th indicator, is the average value of the j-th indicator.

[0223] S306-12, calculate the coefficient of variation v j :

[0224]

[0225] S306-13, calculate the comprehensive information amount c j :

[0226]

[0227] wherein c j is the comprehensive information amount of the j-th indicator, c j is larger, indicating that the j-th indicator covers a larger amount of information; r tj represents the Pearson correlation coefficient of the t-th indicator and the j-th indicator.

[0228] S306-14, calculate the j-th indicator weight ω j :

[0229]

[0230] In the above step S306, the application process of the MARICA multi-attribute decision-making method is as follows:

[0231] The MAIRCA method is an ideal solution-based evaluation tool, which obtains the overall deviation of multiple alternative schemes by accumulating the difference between the ideal value and the actual value of each element of the scheme, and sorts the overall deviation, wherein the alternative scheme with the smallest overall deviation value is closest to the ideal scheme and is the best alternative scheme. Compared with multi-attribute decision-making methods such as VIKOR method and TOPSIS method, the MAIRCA method adopts a linear normalization procedure, and the determined deviation value sorting shows better stability under different weight standards. Based on the MAIRCA method, the importance of the wind monitoring point of a high-speed rail strong wind early warning system in the network is evaluated, the weak link of the system is identified, and a scheme for optimizing the layout of the monitoring point is proposed, thereby providing a reference for the construction of the high-speed rail strong wind early warning system.

[0232] The main steps are as follows:

[0233] S306-21, calculate the ideal evaluation matrix T p :

[0234]

[0235] where t pij denotes the ideal evaluation value of the evaluation object i under index j, denotes the preference of the evaluator to the evaluation object i, ω j denotes the weight of the toughness index. The present application is data-driven, and the evaluation is not affected by subjective preference, i.e.

[0236]

[0237] S306-22, calculate the actual evaluation matrix T r :

[0238] T r = [t rij ] m×n (26)

[0239] For benefit-type index t rij :

[0240]

[0241] For cost-type index t rij :

[0242]

[0243] S306-23, calculate the deviation matrix G:

[0244]

[0245] where g ij is the deviation between the actual value and the ideal value of the jth index in the ith evaluation object.

[0246] S306-24, calculate the total deviation value Q of the evaluation object i :

[0247]

[0248] The prior art method only applies the signal correlation method to two adjacent monitoring points, and cannot well identify the weak points along the whole high-speed rail line. In order to identify the independent wind field of medium and small scale, the monitoring points of the whole line are regarded as a correlated whole and the signal correlation method is used. In addition, a large amount of historical data proves that a strong wind event often activates the early warning response of multiple monitoring points along the high-speed rail line, which will lead to an increase in alarm frequency and an excessively long alarm interval, affecting the efficiency of operation. However, this collective response mode also reveals the great potential of monitoring point research from the overall perspective, opening up a new way to improve the overall performance of the strong wind monitoring system and the efficiency of high-speed rail operation.

[0249] The following is a specific experiment of a high-speed rail line wind speed monitoring point layout optimization method:

[0250] A certain high-speed rail line is located in a temperate monsoon climate zone. The total length of the line is 173.96 km, and the design speed is 120 km / h to 350 km / h. The distribution of the 39 wind speed monitoring points along the line is shown in Figure 4 The initial high-speed rail wind monitoring points are laid out according to meteorological data, terrain reconnaissance, and other data along the line, in accordance with the principles set forth in TB10185-2021 "Technical Specifications for Railway Natural Disaster and Foreign Object Intrusion Monitoring System Engineering" (hereinafter referred to as "the Specification").

[0251] The wind speed along the line shows a clear seasonal characteristic, as shown in Table 1. To mitigate the impact of seasonal changes on the correlation of monitoring points and ensure the reliability of the constructed monitoring point correlation network, this embodiment selects the data of the entire year of a certain high-speed rail line as the basis data, collecting data once every second, with a total of 1,229,904,000 wind speed data points for the entire year.

[0252] Table 1: Statistical Table of Average Wind Speed Along a Certain High-Speed Railway

[0253]

[0254] This embodiment performs preliminary processing on the initial data. First, data containing missing values and significant anomalies (wind speed less than 0 m / s) are removed, and only time steps with records from all monitoring points are retained. Secondly, to maintain the correlation between monitoring points and improve computational efficiency, the data is averaged. Research shows that the average difference between the maximum and minimum wind speeds increases with the length of the time window, but the growth tends to slow down after the window reaches 10 minutes, proving that the 10-minute average can effectively reflect the actual wind speed conditions. Finally, based on the annual data, this embodiment takes the 10-minute average wind speed data of the 39 wind monitoring points, and after cleaning and alignment, 1,640,925 data points are obtained for research. The cleaned wind speed data visualization is shown in Figure 5 .

[0255] Correlation weight determination based on entropy weight method: This embodiment uses the preprocessed data to perform correlation tests between all monitoring points along the high-speed rail line. The Pearson correlation coefficient, Spearman rank correlation coefficient, Kendall rank correlation coefficient, cosine similarity, Euclidean distance, and Canberra distance of the wind speed series are calculated. Due to the complexity of multiple network parallel analysis, the entropy weight method is innovatively introduced to calculate the information entropy weight of each correlation to realize index fusion and complete the reconstruction of the original network topology. The six types of correlation coefficients of the 741 monitoring points formed by the combination of the 39 monitoring points are taken as input data, and the entropy weight method is used to determine the weight values of the six types of evaluation indexes. The normalized data is shown in Table 2.

[0256] Table 2: Normalized index set data

[0257]

[0258] The entropy weight method reflects the importance of various correlation indicators. The information entropy values E determined using equations (16) to (20) j and the weight values W j are shown in Table 3.

[0259] Table 3: Information entropy values and weight values of indicators

[0260]

[0261] Among them, the Euclidean distance and the Canberra distance have the highest weight, indicating that these two distance measures are the most important in reflecting the differences in wind speed data. Distance measures can intuitively quantify the differences between wind speed data, and are particularly suitable for detecting the magnitude and trend of wind speed changes. Second, the Pearson correlation coefficient and the cosine similarity also have high weights, showing the importance of linear correlation and similarity in wind speed sequence analysis. The weight of the Spearman rank correlation coefficient and the Kendall correlation coefficient is relatively low, indicating that the influence of non-linear correlation and rank correlation in wind speed sequence analysis is smaller. This shows that the main relationship between wind speed sequences tends to be linear and directional consistency, rather than complex non-linear or rank relationships.

[0262] Network correction: In the process of correlation testing, this embodiment does not set a threshold to filter non-significant connections to avoid information loss, which leads to a large number of redundant edges in the network, as shown by a in Figure 6 It can be seen that each node in the current network is connected to the other 38 nodes. However, such redundant associations have a negative impact on identifying important nodes in the network.

[0263] Therefore, this embodiment deletes edges in order from low to high according to the association weight by conducting sensitivity analysis on the structural indicators of the network (clustering coefficient, network efficiency, average degree, and number of connected components). According to the mutation points of each indicator in the sensitivity analysis results, the threshold is determined to remove redundant edges in the network. Sensitivity analysis is shown by b in Figure 6

[0264] (1) The clustering coefficient starts to decrease significantly at an edge weight threshold of about 0.5 and tends to zero at about 0.8, which indicates that deleting edges in this interval significantly destroys the local structure with strong association in the network.

[0265] ​(2) Network efficiency starts to decrease significantly at the edge weight threshold of about 0.5, and decreases to a lower level at about 0.8, indicating that in this interval, the average distance between nodes increases significantly, the correlation of the whole network weakens, the information transmission efficiency decreases, and the overall connectivity of the network is destroyed.

[0266] (3) The average degree slowly decreases in the process of deleting edges, but the speed of decrease accelerates at about 0.6, which indicates that at this threshold, the deleted edges begin to significantly reduce the connection number of nodes, further weakening the connectivity of the network.

[0267] (4) The number of connected components increases sharply at the edge weight threshold of about 0.6, and rapidly increases to the maximum value thereafter, showing that the network starts to split into multiple isolated subgraphs in this interval, significantly affecting the overall connectivity of the network.

[0268] Based on the above analysis, the starting point of significant change in network structure and the turning point of network from connected state to split state are both 0.6, so it is taken as the threshold to delete redundant edges with connection weight less than 0.6, and to retain key edges that have greater impact on network structure and function, thereby ensuring the accuracy and reliability of network analysis.

[0269] In the process of setting the threshold to 0.6 to remove invalid edges, it is found that all the connections of the monitoring point ZD5 are deleted, which indicates that the correlation between the wind speed of this monitoring point and the wind speed collected by other monitoring points in the network is low. This indicates that the local wind conditions around this monitoring point are significantly different from the wind conditions observed by the entire monitoring network. In order to more accurately evaluate the wind conditions in this area, it is suggested to add monitoring points at the green point positions as shown in Figure 7 , to strengthen the monitoring of wind speed around this point.

[0270] Modularity analysis of the network:

[0271] The modularity of the network is 0.157, the number of community spaces is 3, and the network presents a sparse connection, but still exhibits a clear community structure. The wind speed is unstable, and it can be known from Figure 2 that the monitoring points are arranged along the railway line with an interval of about 5-10 kilometers, which leads to loose connections based on the correlation of wind speed series, and the network structure is sparse. In addition, the terrain along the railway is complex, and the three community spaces are blocked by mountain barriers, resulting in significant differences in wind speed. Some wind speed monitoring points are close in space and located in the same geographical unit, and the recorded wind speed series shows high correlation, so the network can form a stable community structure, Figure 8 The location of each monitoring point is displayed, with colors representing the community space to which it belongs. The size of the monitoring point is determined by its degree. Based on the analysis of the surrounding terrain, the modular segmentation results based on wind speed sequence correlation are basically consistent with the natural terrain segmentation, and monitoring points in the same community are basically adjacent in space.

[0272] Further analysis of satellite remote sensing images revealed that the two monitoring points, which exhibited differences in community affiliation and spatial proximity, both displayed significant characteristics in a small-scale environment: ① Figure 9 As shown in a, monitoring point ZD13 is surrounded by mountains, forming an independent local wind field with low correlation to the wind conditions of nearby monitoring points. Furthermore, the entire route is affected by monsoon winds, therefore ZD13 was incorrectly classified as community 3. ② As... Figure 9 As shown in b, monitoring point ZD31 is located in a plain area, but its southwest side is blocked by tall buildings, affecting its correlation with wind conditions of most monitoring points along the entire line. This terrain blocking effect is similar to the terrain of Community 2, where the southwest side is blocked by mountains, causing ZD31 to be incorrectly classified as Community 2. Based on the above analysis, the characteristics of this correlation network are consistent with the actual physical world, indicating that the network has been effectively constructed.

[0273] Node centrality index weight determination based on CRITIC method: Based on the construction of an effective network, according to the characteristic analysis of the nodes in the network, six types of indicators describing the importance of nodes in the network are determined as evaluation indicators: degree centrality (DC), betweenness centrality (BC), tight centrality (EC), eigenvector centrality (CC), Page Rank centrality (PR), and K-shell centrality (KS). The index data of each node in the network are shown in Table 4.

[0274] Table 4 Node Centrality Index Data

[0275]

[0276] The CRITIC method, which employs an improved coefficient of variation, integrates the coefficient of variation v of variation that includes correlation information between heterogeneous indicators and the volatility of the indicators themselves. j As the amount of comprehensive information c j The important factors are to determine the objective weights of each centrality index, and the coefficient of variation v is determined according to equations (21) to (24). j Comprehensive information volume c j and indicator weight ω j As shown in Table 5.

[0277] Table 5. Information Entropy Values ​​and Weights of Indicators

[0278]

[0279] Betweenness centrality has the highest weight (weight 0.378), indicating that some nodes play a key role in the connection path. The importance of degree centrality (0.189) and eigenvector centrality (0.173) is also high, indicating that the number of connections of the node and its connection with important nodes have a certain influence. The medium weight of K-shell centrality (0.112) indicates that the hierarchy and stability of core nodes have a certain effect in the network. The weights of closeness centrality (0.0735) and Page Rank centrality (0.0743) are low, indicating that the average correlation distance between nodes and the overall importance in the network are not the most critical factors.

[0280] Result analysis and optimization scheme: Using node centrality index data and index weight, using MARICA multi-attribute decision method, using formula (25) to calculate ideal evaluation matrix Tp, using formula (27) to calculate the actual evaluation value t of the 6 types of indexes in the evaluation matrix E rij , and using formula (26) to construct the actual evaluation matrix Tr. Using formula (29), formula (30), the total deviation value Q of the evaluation object is calculated by the ideal evaluation matrix Tp and the actual evaluation matrix Tr i . In this embodiment, the correlation between the monitoring points has a positive meaning for monitoring, so according to formula (28), the priority ranking of each monitoring point is calculated as shown in Table 6.

[0281] Table 6 Deviation value and ranking

[0282]

[0283] This case verifies the applicability of the proposed method through example analysis. For example, ZD36>ZD35>ZD37, it is suggested to add a monitoring point between ZD36 and ZD37. Since ZD34-ZD39 is not continuous with other monitoring points, the decision rule is used separately in the decision of the new site. In summary, a total of 13 new monitoring points are needed in the whole line, and the specific interval is: ZD3-ZD4, ZD4-ZD5, ZD5-ZD6, ZD13-ZD14, ZD14-ZD15, ZD18-ZD19, ZD30-ZD31, ZD31-ZD32, ZD36-ZD37.

[0284] In actual engineering, the goal of wind monitoring point layout optimization changes with demand. For example, to enhance the sensing ability of the monitoring network to strong wind events, the monitoring points can be added in order of importance from high to low according to the importance of key nodes; if the goal is to reduce economic cost, then under the premise of meeting the "specification", the monitoring points with lower importance can be deleted first.

[0285] The embodiment also provides a high-speed rail line wind speed monitoring point layout optimization device, which is used for realizing the above-mentioned embodiment and preferred embodiment, and has been described above. As used below, the terms "module", "unit", "sub-unit", and the like can be a combination of software and / or hardware that realizes a predetermined function. Although the device described in the following embodiment is preferably realized in software, realization in hardware, or a combination of software and hardware is also possible and contemplated.

[0286] Figure 10 is a structural block diagram of a high-speed rail line wind speed monitoring point layout optimization device according to the embodiment of the application, as shown in Figure 10 , the device comprises:

[0287] The data acquisition module 10 is configured to acquire wind speed data of each monitoring point in an initial layout position; and generate a multi-dimensional correlation index set based on the wind speed data between each pair of monitoring points.

[0288] The network construction module 20 is configured to perform weight distribution on each index based on the multi-dimensional correlation index set, generate a weight distribution result, and calculate a correlation weight between each pair of monitoring points; and construct an initial correlation network for the monitoring points based on the weight distribution result and the correlation weight.

[0289] The network correction module 30 is configured to confirm, among all connections of the initial correlation network, a connection with a correlation weight lower than a preset weight threshold as a weak connection, delete the weak connection, and form a corrected target correlation network; and adjust the initial layout position of the monitoring points to an optimized layout position based on the target correlation network.

[0290] The data acquisition module 10 is further configured to calculate a linear correlation coefficient, a nonlinear rank correlation coefficient, a similarity, and a distance metric index between each pair of monitoring points, respectively, and generate the multi-dimensional correlation index set based on the linear correlation coefficient, the nonlinear rank correlation coefficient, the similarity, and the distance metric index.

[0291] The network construction module 20 is further configured to perform standardization processing on each index in the multi-dimensional correlation index set, calculate an information entropy and an information utility value corresponding to each index, and determine a weight distribution result of each index based on the information utility value.

[0292] The network correction module 30 is further configured to determine the preset weight threshold based on an acquired network structure index, and the network structure index comprises a clustering coefficient, a network efficiency, an average degree, and a number of connected components of the initial correlation network.

[0293] The network correction module 30 is further configured to calculate a monitoring point priority ranking result for each monitoring point of the target associated network; and adjust the initial layout position of the monitoring point to an optimized layout position based on the target associated network and the monitoring point priority ranking result.

[0294] The network correction module 30 is further configured to calculate a multi-dimensional node importance evaluation index for each monitoring point of the target associated network.

[0295] The network correction module 30 is further configured to calculate a node index actual value based on the multi-dimensional node importance evaluation index.

[0296] The network correction module 30 is further configured to determine the monitoring point priority ranking result by calculating a total deviation value of the node index actual value and a preset index ideal value.

[0297] The network correction module 30 is further configured to calculate a coefficient of variation based on the wind speed data; and determine an importance index weight of the multi-dimensional node importance evaluation index based on the multi-dimensional correlation index set and the coefficient of variation.

[0298] The network correction module 30 is further configured to calculate the node index actual value based on the multi-dimensional node importance evaluation index and the importance index weight.

[0299] The network correction module 30 is further configured to perform module division on the target associated network to obtain an associated network module; and match the associated network module with a terrain feature based on the obtained terrain feature of the monitoring point to obtain a terrain feature verification result.

[0300] The network correction module 30 is further configured to adjust the initial layout position of the monitoring point to the optimized layout position based on the target associated network, the monitoring point priority ranking result, and the terrain feature verification result.

[0301] It should be noted that each of the above modules can be a functional module or a program module, and can be implemented by software or hardware. For the modules implemented by hardware, each of the above modules can be located in the same processor; or each of the above modules can be located in different processors in any combination.

[0302] In addition, in combination with the high-speed rail line wind speed monitoring point layout optimization method in the above embodiment, the embodiment of the application can provide a storage medium for implementation. The storage medium has a computer program stored thereon; and the computer program is executed by a processor to implement any one of the high-speed rail line wind speed monitoring point layout optimization methods in the above embodiments.

[0303] Those skilled in the art should understand that each technical feature of the above-described embodiments can be combined arbitrarily, and for the sake of brevity, each technical feature of the above-described embodiments is not described in all possible combinations, however, as long as the combinations of the technical features do not exist, it should be considered that it is within the scope of the description.

[0304] The above-described embodiments only express several implementation manners of the present application, the description is more specific and detailed, but it should not be understood as a limitation on the patent scope of the application. It should be pointed out that for ordinary skilled in the art, without departing from the concept of the present application, a number of modifications and improvements can be made, which are within the protection scope of the present application. Therefore, the protection scope of the patent of the present application should be subject to the appended claims.< / k> < / k> < / k> < / k>

Claims

1. A high-speed rail line wind speed monitoring point layout optimization method, characterized in that, The method comprises the following steps: acquiring wind speed data of each monitoring point in an initial layout position; generating a multi-dimensional correlation index set based on the wind speed data between each pair of monitoring points; weighting each index based on the multi-dimensional correlation index set, generating a weight distribution result, and calculating the correlation weight between each pair of monitoring points; constructing an initial correlation network for the monitoring points based on the weight distribution result and the correlation weight; confirming the connection with a correlation weight lower than a preset weight threshold as a weak connection in all connections of the initial correlation network, deleting the weak connection, and forming a modified target correlation network; adjusting the initial layout position of the monitoring points to an optimized layout position based on the target correlation network.

2. The high-speed railway line along wind speed monitoring point layout optimization method according to claim 1, characterized in that, The method comprises the following steps: calculating a linear correlation coefficient, a nonlinear rank correlation coefficient, a similarity and a distance metric index for the wind speed data between each pair of monitoring points, and generating the multi-dimensional correlation index set based on the linear correlation coefficient, the nonlinear rank correlation coefficient, the similarity and the distance metric index.

3. The method according to claim 1, wherein, The method comprises the following steps: standardizing each index in the multi-dimensional correlation index set, calculating the information entropy and information utility value corresponding to each index, and determining the weight distribution result of each index based on the information utility value.

4. The method according to claim 1, wherein, The preset weight threshold is determined according to an acquired network structure index, and the network structure index includes the clustering coefficient, network efficiency, average degree and connected component number of the initial correlation network.

5. The method of claim 1, wherein, The method comprises the following steps: calculating a monitoring point priority ranking result for each monitoring point of the target correlation network; and adjusting the initial layout position of the monitoring points to an optimized layout position based on the target correlation network and the monitoring point priority ranking result.

6. The method of claim 5, wherein, The method comprises the following steps: calculating a multi-dimensional node importance evaluation index for each monitoring point of the target correlation network; calculating a node index actual value based on the multi-dimensional node importance evaluation index; determining the monitoring point priority ranking result by calculating the total deviation value of the node index actual value and a preset index ideal value.

7. The method of claim 6, wherein, The method comprises the following steps: calculating a coefficient of variation based on the wind speed data; determining an importance index weight for the multi-dimensional node importance evaluation index based on the multi-dimensional correlation index set and the coefficient of variation; calculating a node index actual value based on the multi-dimensional node importance evaluation index and the importance index weight.

8. The method of claim 5, wherein, The adjusting the initial layout position of the monitoring points to an optimized layout position based on the target correlation network and the monitoring point priority ranking result comprises: performing module division on the target correlation network to obtain a correlation network module, and matching the correlation network module with the terrain features based on the obtained terrain features of the monitoring points to obtain a terrain feature verification result; adjusting the initial layout position of the monitoring points to an optimized layout position based on the target correlation network, the monitoring point priority ranking result, and the terrain feature verification result.

9. A device for optimizing the layout of wind speed monitoring points along a high-speed railway, characterized in that, comprises: a data acquisition module configured to acquire wind speed data of each monitoring point in an initial layout position; generate a multi-dimensional correlation index set based on the wind speed data between each pair of monitoring points; a network construction module configured to perform weight distribution on each index based on the multi-dimensional correlation index set to generate a weight distribution result, and calculate the correlation weight between each pair of monitoring points; construct an initial correlation network for the monitoring points based on the weight distribution result and the correlation weight; a network correction module configured to, in all connections of the initial correlation network, identify connections with a correlation weight lower than a preset weight threshold as weak connections, delete the weak connections, and form a corrected target correlation network; adjust the initial layout position of the monitoring points to an optimized layout position based on the target correlation network.

10. A storage medium, characterized by The storage medium stores a computer program, wherein the computer program is configured to execute the high-speed rail line wind speed monitoring point layout optimization method of any one of claims 1-8 when running.

Images