A communication tower risk monitoring method and system based on deep learning

Abstract

The present application relates to the technical field of communication tower risk monitoring, and in particular to a communication tower risk monitoring method and system based on deep learning. The method includes: obtaining a structural static parameter set, and determining a structural feature information set based on the structural static parameter set; obtaining an environmental factor information set, analyzing the environmental factor information set, and determining a dynamic change feature information set; based on the structural feature information set and the dynamic change feature information set, fusion analysis is performed on the structural feature information set and the dynamic change feature information set to construct a tilt risk prediction model; based on the tilt risk prediction model, the tilt risk change information is determined and output. The present application realizes the scientific prediction of the tilt risk of railway communication towers, and provides the corresponding tilt risk change information to the railway communication tower maintenance personnel, effectively avoiding the short-term rapid progression of the tilt risk of railway communication towers caused by the lag in maintenance work, and effectively ensuring that the tilt risk of railway communication towers can be eliminated in a timely manner.

Description

Technical Field

[0001] The present application relates to the technical field of communication tower risk monitoring, and in particular to a communication tower risk monitoring method and system based on deep learning. Background Art

[0002] Railway communication towers, as an important part of the railway communication system, are responsible for transmitting key information and signals. Their structural safety directly affects the safety and efficiency of railway transportation. Therefore, it is crucial to ensure the stability and safety of communication towers.

[0003] The existing railway communication tower tilt risk monitoring method relies on the monitoring data provided by the wireless sensors installed on the railway communication tower, which can directly prove that the railway communication tower is tilted. Although this method can reflect the tilt state of the railway communication tower, it is difficult to avoid the short-term rapid progression of the railway communication tower tilt risk caused by the lag in maintenance work, and cannot effectively ensure that the railway communication tower tilt risk is eliminated in time. Summary of the invention

[0004] The present application provides a communication tower risk monitoring method and system based on deep learning to solve the above technical problems.

[0005] In a first aspect, the present application provides a communication tower risk monitoring method based on deep learning, the method comprising:

[0006] Acquire a structural static parameter set, and determine a structural feature information set according to the structural static parameter set;

[0007] Acquire an environmental factor information set, analyze the environmental factor information set, and determine a dynamic change characteristic information set;

[0008] According to the structural feature information set and the dynamic change feature information set, a fusion analysis is performed on the structural feature information set and the dynamic change feature information set to construct a tilt risk prediction model;

[0009] According to the tilt risk prediction model, tilt risk change information is determined and output.

[0010] Through this scheme, based on the structural static parameter set, the structural characteristic information set reflecting the important structural characteristics of the railway communication tower body is analyzed, and by analyzing the environmental factor information set, the dynamic change characteristic information set corresponding to the dynamic environmental factors from the outside that have a direct impact on the tilt risk of the railway communication tower is determined. Through the fusion analysis of the structural characteristic information set and the dynamic change characteristic information set, a tilt risk prediction model reflecting the mapping relationship between structural characteristics and dynamic environmental factors and tilt risks is obtained, so as to obtain the corresponding tilt risks of the railway communication tower at different time points, realize the scientific prediction of the tilt risk of the railway communication tower, and provide the corresponding tilt risk change information to the railway communication tower maintenance personnel, effectively avoiding the short-term rapid progression of the tilt risk of the railway communication tower caused by the lag of maintenance work, and effectively ensuring that the tilt risk of the railway communication tower can be eliminated in time.

[0011] Optionally, the structural static parameter set includes communication tower structure scanning information, communication tower installation information and communication tower material information, and determining the structural feature information set according to the structural static parameter set includes:

[0012] Determine a connection node position set and a connection component angle set according to the communication tower structure scanning information;

[0013] Constructing an undirected graph of communication tower structures according to the connection node position set and the connection component angle set;

[0014] Analyze the communication tower installation information, and extract the key installation node depth and installation node relative distance set corresponding to each installation node;

[0015] Based on the material information of the communication tower, the connection node position set and the connection component angle set are analyzed to determine the connection component strength between any two connected connection nodes, thereby constructing a connection component strength set corresponding to each connection node;

[0016] Based on the communication tower structure undirected graph, constructing a set of structural static feature vectors according to the key installation node depth, the installation node relative distance set and the connection component strength set;

[0017] The structural static feature vector set is used as the structural feature information set.

[0018] Through this solution, a systematic analysis of the connection nodes, connection component structure, installation information and material properties of the communication tower was carried out, laying a solid foundation for the construction of the tilt prediction model. It not only enhanced the understanding of the physical characteristics of the communication tower, but also provided relevant features for the subsequent application of more complex algorithms, significantly improving the accuracy and reliability of subsequent tilt predictions, and providing a basis for preventing and controlling the tilt problem of railway communication towers.

[0019] Optionally, constructing a communication tower structure undirected graph according to the connection node position set and the connection component angle set includes:

[0020] Analyze the connection node position set and the connection component angle set to determine the number of connection nodes and the connection relationship pointing value between any two connection nodes;

[0021] According to the number of connection nodes and the connection relationship, a connection relationship matrix is ​​constructed, which is specifically the following expression:

[0022] ;

[0023] in, is the connection relationship matrix, is the number of connected nodes, Representative The connection node and The connection relationship between the connection nodes points to a value;

[0024] Based on the connection relationship matrix, the communication tower structure undirected graph is constructed according to the connection node position set and the connection component angle set.

[0025] Through this scheme, by analyzing the connection node position set and the connection component angle set, the number of connection nodes and the connection relationship pointing value can be accurately determined, and a connection relationship matrix for quantifying the relationship between each connection node in the railway communication tower is constructed to construct a communication tower structure undirected graph. The connection relationship matrix is ​​used to provide a clear relationship pointing data set for the construction of the communication tower structure undirected graph, so that the communication tower structure undirected graph can be automatically and quickly constructed according to the connection relationship matrix.

[0026] Optionally, the constructing of a structural static feature vector set based on the communication tower structure undirected graph according to the key installation node depth, the installation node relative distance set and the connection component strength set includes:

[0027] Extracting the spatial coordinates of each connection node according to the communication tower structure undirected graph;

[0028] Based on the spatial coordinates, according to the key installation node depth, the installation node relative distance set and the connection component strength set, the structural feature vector corresponding to each connection node is determined, and a structural feature matrix is ​​constructed based on this, which is specifically expressed as follows:

[0029] ;

[0030] in, is the structural feature matrix, For the The structural feature vectors of the connected nodes, For the The horizontal coordinate of the connection node, For the The vertical coordinate of the connection node, For the The height coordinates of the connection nodes; For the The key installation node depth of the connection node. If the current connection node is not the installation node, this item is 0; For the The relative distance set of the installation nodes corresponding to the connection nodes. If the current connection node is not the installation node, this item is 0; For the A set of connection component strengths corresponding to each of the connection nodes;

[0031] Based on the connection relationship matrix and the structural feature matrix, the structural aggregation feature vector of each connection node is determined, and the structural static feature vector set is constructed based on this.

[0032] Through this scheme, the undirected graph of the communication tower structure is analyzed, and the spatial coordinates corresponding to each connection node are extracted. On this basis, the key installation node depth, the relative distance set of the installation nodes and the connection component strength set corresponding to each connection node are combined, and a structural feature matrix is ​​constructed through mathematical analysis. The characteristics of each connection node are integrated and described using mathematical language to facilitate subsequent quantitative analysis, performance simulation and stability prediction. Further, based on the connection relationship matrix and the structural feature matrix, a structural aggregation feature vector is constructed that can reflect the stability performance of the structural characteristics of the connection node under the influence of different connection relationships.

[0033] Optionally, determining the structural aggregation feature vector of each connection node based on the connection relationship matrix and according to the structural feature matrix includes:

[0034] Determine the number of connection paths corresponding to each connection node according to the connection component strength set corresponding to each connection node;

[0035] Based on the number of connection paths, according to the connection relationship matrix and the structural feature matrix, the structural aggregation feature vector of each connection node is determined, specifically as follows:

[0036] ;

[0037] in, For the The aggregate feature vector corresponding to the connected nodes, For the The number of connection paths corresponding to the connection nodes, is the number of connected nodes, For the The connection node and The connection relationship between the connection nodes points to a value, For the The structural feature vectors of the connection nodes.

[0038] Through this scheme, by means of mathematical analysis, on the basis of the number of connection paths, according to the connection relationship matrix and the structural feature matrix, the corresponding mathematical formula is designed to achieve accurate and scientific quantification of the aggregated feature vectors of the connection nodes. By introducing the aggregated feature vectors, the interaction between the information of the connection nodes and their environment can be more comprehensively captured, thereby improving the accuracy of subsequent tilt risk prediction.

[0039] Optionally, the environmental factor information set includes meteorological change information and geological change information, and the step of analyzing the environmental factor information set to determine the dynamic change characteristic information set includes:

[0040] Extracting wind force change information, temperature change information and meteorological carrier change information according to the meteorological change information;

[0041] Extracting soil bearing change information, vibration conduction change information and settlement change information according to the geological change information;

[0042] The dynamic change characteristic information set is constructed based on the wind force change information, temperature change information, meteorological carrier change information, soil bearing change information, vibration conduction change information and settlement change information.

[0043] Through this scheme, the dynamic environmental factors that affect the risk of railway communication tower tilt are divided into two dimensions according to the spatial dimension, and the wind change information, temperature change information, meteorological carrier change information, soil bearing change information, vibration conduction change information and settlement change information in the meteorological change information and geological change information are extracted respectively. Based on this, a dynamic change feature information set is constructed to provide scientific data support for the subsequent all-round prediction of the risk of communication tower tilt.

[0044] Optionally, performing a fusion analysis on the structural feature information set and the dynamic change feature information set according to the structural feature information set and the dynamic change feature information set to construct a tilt risk prediction model, including:

[0045] By using a preset recurrent neural network model, a time series analysis is performed on each dynamic change feature in the dynamic change feature information to determine a dynamic time series feature vector set;

[0046] Based on the dynamic time series feature vector set and according to the structural static feature vector set, determining the fusion influence factor of different dynamic features at different time points on the structural feature vector corresponding to each connection node;

[0047] Based on the structural static feature vector set and according to the fusion influencing factors, the comprehensive tilt risk at different time points is determined, and the tilt risk prediction model is constructed based on this.

[0048] Through this scheme, a time series analysis is performed on each dynamic change feature in the dynamic change feature information to determine a dynamic time series feature vector set, so as to capture the dynamic changes of the railway communication tower affected by time and external environmental factors during operation. On the basis of the dynamic time series feature vector set, combined with the structural static feature vector set, a structural fusion influencing factor that can reflect the degree of influence of different dynamic features on the static structure in different time bags is determined, and then the comprehensive tilt risk at different time points is determined, so as to construct a tilt risk prediction model that can predict the tilt risk of the railway communication tower within a sufficient time margin.

[0049] Optionally, based on the dynamic time series feature vector set, according to the structural static feature vector set, determining the fusion influence factor of different dynamic features at different time points on the structural feature vector corresponding to each connection node is specifically the following formula:

[0050] ;

[0051] in, is the fusion impact factor, is the preset ratio conversion coefficient, For the The structural feature vectors of the connection nodes, is the first Dynamic features at time points The corresponding vector value under For the current The corresponding preset feature influence coefficient.

[0052] Through this scheme, mathematical analysis methods are used to design mathematical formulas based on the dynamic time series feature vector set and the structural static feature vector set. The influence of different dynamic features on the current connection point at different time points is combined and analyzed, and the comprehensive impact of environmental changes on the connection nodes is captured. The fusion impact factor is quantified to improve the accuracy of the prediction of the risk of railway communication tower tilt.

[0053] Optionally, based on the structural static feature vector set and according to the fusion impact factor, the comprehensive tilt risk at different time points is determined, specifically by the following formula:

[0054] ;

[0055] in, For time point The corresponding comprehensive tilt risk is as follows: is the preset risk conversion factor, is the number of connected nodes, For the The structural feature vectors of the connection nodes, is the fusion influencing factor.

[0056] Through this scheme, mathematical analysis methods are used to design mathematical formulas based on the set of structural static feature vectors and fusion influencing factors. The contribution of each connection point to the overall tilt risk of the railway communication tower under the influence of dynamic characteristics is superimposed, and then the local characteristic changes of each connection node are summarized into an overall risk assessment, so as to quantify the comprehensive tilt risk that can standardize the overall tilt risk of the railway communication tower, and significantly improve the prediction accuracy of the tilt risk of the railway communication tower.

[0057] In a second aspect, the present application provides a communication tower risk monitoring system based on deep learning, the system comprising:

[0058] A structural analysis module, used to obtain a structural static parameter set, and determine a structural feature information set according to the structural static parameter set;

[0059] A dynamic feature analysis module, used to obtain an environmental factor information set, analyze the environmental factor information set, and determine a dynamic change feature information set;

[0060] A model building module, used to perform fusion optimization analysis on the structural feature information set and the dynamic change feature information set according to the structural feature information set and the dynamic change feature information set, so as to build a tilt risk prediction model;

[0061] The risk output module is used to determine and output tilt risk change information according to the tilt risk prediction model. BRIEF DESCRIPTION OF THE DRAWINGS

[0062] In order to more clearly illustrate the embodiments of the present application or the technical solutions in the prior art, a brief introduction will be given below to the drawings required for use in the embodiments or the description of the prior art. Obviously, the drawings described below are some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying any creative labor.

[0063] Figure 1 A schematic diagram of an application scenario provided for an embodiment of the present application;

[0064] Figure 2 A flowchart of a communication tower risk monitoring method based on deep learning provided in one embodiment of the present application;

[0065] Figure 3 A schematic diagram of the structure of a communication tower risk monitoring system based on deep learning provided in one embodiment of the present application. DETAILED DESCRIPTION

[0066] In order to make the purpose, technical scheme and advantages of the embodiments of the present application clearer, the technical scheme in the embodiments of the present application will be clearly and completely described below in conjunction with the drawings in the embodiments of the present application. Obviously, the described embodiments are part of the embodiments of the present application, not all of the embodiments. Based on the embodiments in the present application, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of this application.

[0067] In addition, the term "and / or" in this article is only a description of the association relationship of associated objects, indicating that there can be three relationships. For example, A and / or B can represent: A exists alone, A and B exist at the same time, and B exists alone. In addition, the character " / " in this article, unless otherwise specified, generally means that the associated objects before and after are in an "or" relationship.

[0068] The embodiments of the present application are further described in detail below in conjunction with the drawings in the specification.

[0069] The existing railway communication tower tilt risk monitoring method relies on the monitoring data provided by the wireless sensors installed on the railway communication tower, which can directly prove that the railway communication tower is tilted. Although this method can reflect the tilt state of the railway communication tower, it is difficult to avoid the short-term rapid progression of the railway communication tower tilt risk caused by the lag in maintenance work, and cannot effectively ensure that the railway communication tower tilt risk is eliminated in time.

[0070] Based on this, the present application provides a communication tower risk monitoring method and system based on deep learning. Through the fusion analysis of the structural feature information set and the dynamic change feature information set, a tilt risk prediction model reflecting the mapping relationship between the structural features and dynamic environmental factors and the tilt risk is obtained, so as to obtain the corresponding tilt risk of the railway communication tower at different time points, realize the scientific prediction of the tilt risk of the railway communication tower, and provide the corresponding tilt risk change information to the railway communication tower maintenance personnel, effectively avoiding the short-term rapid progression of the railway communication tower tilt risk caused by the lag of maintenance work, and effectively ensuring that the tilt risk of the railway communication tower can be eliminated in time.

[0071] Figure 1 A schematic diagram of an application scenario provided by this application. In the process of analyzing the risk of railway communication tower tilt, the method provided by this application is applied to achieve scientific prediction of the risk of railway communication tower tilt.

[0072] Specifically, the method of the present application is applied to any server, which communicates with a number of wireless environmental parameter monitoring sensors, and obtains and analyzes a structural static parameter set provided by a railway communication tower maintenance personnel through the server. Based on the structural static parameter set, a structural feature information set reflecting the important structural features of the railway communication tower body is analyzed, and by analyzing the environmental factor information set provided by a number of wireless environmental parameter monitoring sensors, a dynamic change feature information set corresponding to the dynamic environmental factors from the outside that have a direct impact on the tilt risk of the railway communication tower is determined. Through the fusion analysis of the structural feature information set and the dynamic change feature information set, a tilt risk prediction model that reflects the mapping relationship between the structural characteristics and the dynamic environmental factors and the tilt risk is obtained, so as to obtain the corresponding tilt risk of the railway communication tower at different time points, realize the scientific prediction of the tilt risk of the railway communication tower, and provide the corresponding tilt risk change information to the railway communication tower maintenance personnel, effectively avoiding the short-term rapid progression of the tilt risk of the railway communication tower caused by the lag in maintenance work, and effectively ensuring that the tilt risk of the railway communication tower can be eliminated in time.

[0073] For specific implementation methods, please refer to the following embodiments.

[0074] Figure 2 This is a flowchart of a communication tower risk monitoring method based on deep learning provided in one embodiment of the present application. The method of this embodiment can be applied to the server in the above scenario. Figure 2 As shown, the method includes:

[0075] S201. Obtain a structural static parameter set, and determine a structural feature information set according to the structural static parameter set.

[0076] The structural static parameter set may be a set of fixed parameters related to the physical structure of the railway communication tower, such as height, material, foundation type, component geometry, etc. The structural static parameter set may be provided by railway communication tower maintenance personnel.

[0077] The structural characteristic information set may be a set of important characteristic information about the tower structure derived from the static parameter set, such as bearing capacity, structural strength, etc.

[0078] Specifically, the tilt risk of railway communication towers is mainly affected by their own structures and external dynamic factors. The analysis of the static structural parameters of the railway communication tower itself can reflect the static stability of the tower, provide a benchmark data framework for the introduction and analysis of the influence of external dynamic factors, and at the same time improve the adaptability of the tilt risk prediction process to towers with different structures. Through mathematical analysis, a series of parameter characteristics reflecting the structure of the communication tower in the structural static parameter set are described in mathematical language to achieve the unification of mathematical processing dimensions and provide a data framework for subsequent tilt risk analysis.

[0079] S202: Acquire an environmental factor information set, analyze the environmental factor information set, and determine a dynamic change feature information set.

[0080] The environmental factor information set may be an information set of external dynamic conditions that affect the stability of railway communication towers, such as dynamic environmental parameters such as temperature, wind speed, and soil quality. The environmental factor information set may be obtained by several wireless environmental parameter monitoring sensors of different types deployed around railway communication towers.

[0081] The dynamically changing feature information set may be a feature set about the changes of these environmental factors over time obtained after analyzing the environmental factor information set.

[0082] Specifically, external environmental factors have a direct and dynamic impact on the tilt risk of railway communication towers. The environmental factors in the same area show overall periodic changes and are accompanied by sudden extreme changes. With the support of modern environmental monitoring technology, the overall periodic changes of environmental factors can be monitored more accurately and sufficient time margin can be provided. For sudden extreme changes, the time margin provided is significantly shorter. This requires that the prediction of the tilt risk of railway communication towers needs to have the characteristics of strong real-time performance. Through mathematical analysis, the collected environmental factor information is analyzed in time series to obtain a dynamically changing feature information set that reflects the characteristic relationship between environmental factor changes and time changes, so as to ensure the real-time performance of the subsequent tilt risk prediction process that introduces dynamic environmental factors.

[0083] S203: Based on the structural feature information set and the dynamic change feature information set, a fusion analysis is performed on the structural feature information set and the dynamic change feature information set to construct a tilt risk prediction model.

[0084] The tilt risk prediction model can be a mathematical model for predicting the tilt risk change of railway communication towers based on deep learning technology and trained according to a structural feature information set and a dynamic change feature information set.

[0085] Specifically, due to the extremely complex nonlinear relationship between the structural characteristics of railway communication towers, external dynamic environmental factors and the tilt risk of railway communication towers, it is difficult to directly point the structural characteristics and external dynamic environmental factors to the tilt risk of railway communication towers through a clear mathematical relationship. Through deep learning technology, the monitoring data of railway communication towers at different historical time points are subjected to time series fitting analysis to train mathematical factors that tend to reflect the relationship between structural characteristics and dynamic environmental factors and tilt risks. Then, combined with the currently obtained structural characteristics and dynamic environmental factors, the structural characteristic information set and the dynamic change characteristic information set are fused and analyzed to construct a tilt risk prediction model that can map the tilt risk of railway communication towers, so as to achieve scientific prediction of the tilt risk of railway communication towers.

[0086] S204: Determine and output tilt risk change information according to the tilt risk prediction model.

[0087] The tilt risk change information may be the corresponding railway communication tower tilt risk information at different time points obtained based on the tilt risk prediction model.

[0088] Specifically, the tilt risk prediction model is used to map the impact of dynamic environmental factors on the tower structure at different time points to the changing process of the tilt risk, so as to obtain the corresponding railway communication tower tilt risk at different time points. The tilt risk change information is presented in an appropriate visual form, such as a change curve, through data visualization technology, and the corresponding visualization data is provided to the railway communication tower maintenance personnel through human-computer interaction equipment.

[0089] Through this scheme, based on the structural static parameter set, the structural characteristic information set reflecting the important structural characteristics of the railway communication tower body is analyzed, and by analyzing the environmental factor information set, the dynamic change characteristic information set corresponding to the dynamic environmental factors from the outside that have a direct impact on the tilt risk of the railway communication tower is determined. Through the fusion analysis of the structural characteristic information set and the dynamic change characteristic information set, a tilt risk prediction model reflecting the mapping relationship between structural characteristics and dynamic environmental factors and tilt risks is obtained, so as to obtain the corresponding tilt risks of the railway communication tower at different time points, realize the scientific prediction of the tilt risk of the railway communication tower, and provide the corresponding tilt risk change information to the railway communication tower maintenance personnel, effectively avoiding the short-term rapid progression of the tilt risk of the railway communication tower caused by the lag of maintenance work, and effectively ensuring that the tilt risk of the railway communication tower can be eliminated in time.

[0090] In some embodiments, based on the communication tower structure scanning information, the connection node position set and the connection component angle set are determined; based on the connection node position set and the connection component angle set, an undirected graph of the communication tower structure is constructed; the communication tower installation information is analyzed to extract the key installation node depth and the installation node relative distance set corresponding to each installation node; based on the communication tower material information, the connection node position set and the connection component angle set are analyzed to determine the connection component strength between any two connected connection nodes, thereby constructing a connection component strength set corresponding to each connection node; based on the communication tower structure undirected graph, a structural static feature vector set is constructed according to the key installation node depth, the installation node relative distance set and the connection component strength set; the structural static feature vector set is used as the structural feature information set.

[0091] The structural static parameter set includes communication tower structure scanning information, communication tower installation information and communication tower material information.

[0092] The communication tower structure scanning information may be scanning point information obtained by scanning the railway communication tower structure using a scanning device, such as a laser scanner.

[0093] The communication tower installation information may be relevant information about whether there is a hard connection between the railway communication tower structure and other objects or the ground to achieve the communication tower installation.

[0094] The communication tower material information may be material parameter information of various components constituting the communication tower.

[0095] The connection node position set may be a set of spatial position coordinates of all connection nodes in the railway communication tower, and the connection nodes may be connection points of different structural components in the railway communication tower.

[0096] The connection component angle set may be a set of angle information of connection components (such as steel cables, support rods, etc.) between connection nodes in space.

[0097] The communication tower structure undirected graph can be based on the data structure of the undirected graph, abstracting the connection nodes into graph nodes, abstracting the connection components into edges, and then constructing a mathematical model for describing the structural characteristics of the communication tower.

[0098] The installation node may be a part of a special connection node in a railway communication tower used to maintain the installation stability of the railway communication tower.

[0099] The critical mounting node depth may be the depth of the mounting node in a medium (eg, soil) to which a hard connection exists.

[0100] The installation node relative distance set may be a set of relative distances between an installation node and other installation nodes.

[0101] The strength of the connection component can be the force strength of the connection component between the connection points under the influence of materials and installation angles.

[0102] The connection component strength set may be a strength information set corresponding to a connection point and several connection components between other connection points having connection relationships.

[0103] The structural static feature vector set may be a set of vector information that can reflect a series of static structural features of a railway communication tower.

[0104] Specifically, in the process of analyzing the static structure of the railway communication tower, we mainly start from two directions. One is to abstract the geometric structure of the railway communication tower itself, and the other is to analyze its structural stability performance based on this abstract structure. Among them, clarifying the position of the connection node is an important prerequisite for understanding the geometric structure of the communication tower. By analyzing the structural scanning information of the communication tower, the spatial distribution position of each connection node in the communication tower can be extracted, and the angle of the connection component between each connection node can be extracted. The angle of the connection component can not only reflect the local collection result of the communication tower, but also affect the way it bears and transmits stress. Based on the connection node position set and the connection component angle set, the connection node is regarded as a node in an undirected graph, and the connection node position is used as a node attribute, the connection component between the connection nodes is regarded as an edge in the undirected graph, and the connection component angle is used as the attribute of the corresponding edge, so as to construct an undirected graph of the communication tower structure that can reflect the abstract structure of the railway communication tower. In the process of analyzing the structural stability performance, it is necessary to clarify the installation node characteristics of the railway communication tower, so as to introduce external dynamic factors in the future. The key installation node depth and the relative distance of the installation nodes will affect the stability of the tower body and its ability to resist external forces. This information is crucial for evaluating the stability of the tower base. The key installation node depth and the corresponding installation node relative distance set corresponding to each installation node are extracted through the key minutes of the communication tower installation recorded in the communication tower installation information. In the analysis of the strength of the connection components, the strength test data of different connection components at different connection angles are used to fit the influence ratio of different connection angles on the strength, and then the connection component strength is determined according to the current connection component material and the corresponding connection angle reflected in the communication tower material information and the connection component angle set, and then the connection component strength set corresponding to each connection node is constructed. The parameter features corresponding to the undirected graph of the communication tower structure, the key installation node depth, the installation node relative distance set and the connection component strength set are converted into corresponding vector features through a feature vector conversion model, such as a convolutional neural network or a recurrent neural network, to achieve unified processing of data dimensions.

[0105] Through this solution, a systematic analysis of the connection nodes, connection component structure, installation information and material properties of the communication tower was carried out, laying a solid foundation for the construction of the tilt prediction model. It not only enhanced the understanding of the physical characteristics of the communication tower, but also provided relevant features for the subsequent application of more complex algorithms, significantly improving the accuracy and reliability of subsequent tilt predictions, and providing a basis for preventing and controlling the tilt problem of railway communication towers.

[0106] In some embodiments, the connection node position set and the connection component angle set are analyzed to determine the number of connection nodes and the connection relationship pointing value between any two connection nodes; and a connection relationship matrix is ​​constructed according to the number of connection nodes and the connection relationship, which is specifically the following expression (1):

[0107]

[0108] in, is the connection relationship matrix, is the number of connected nodes, Representative The connection node and The connection relationship points to the value between the connection nodes; based on the connection relationship matrix, according to the connection node position set and the connection component angle set, an undirected graph of the communication tower structure is constructed.

[0109] The number of connection nodes may be the total number of connection nodes within the currently analyzed railway communication tower.

[0110] The connection relationship pointing value can be a value used to characterize whether there is a connection relationship between two connection nodes. The connection relationship pointing value can be set as follows: when there is a connection relationship between the two connection nodes, the corresponding connection relationship pointing value is 1; when there is no connection relationship between the two connection nodes, the corresponding connection relationship pointing value is 0.

[0111] The connection relationship matrix may be a mathematical matrix containing connection relationship characteristics between any two connection points in a communication tower.

[0112] Specifically, determining the number of connection nodes can clearly understand the complexity of the structure and conduct a systematic analysis of the subsequent construction of the matrix and undirected graph; the connection relationship orientation value emphasizes the strength of the interaction between the connection nodes, which helps to accurately reflect the impact of different connection designs on the overall stability; based on the number of connection nodes and the connection relationship orientation value, the connection relationship matrix constructed according to mathematical expression (1) quantifies the relationship between the connection nodes in the railway communication tower, providing structured data for subsequent graph theory analysis and algorithm application, which helps to evaluate the connectivity and stability of the entire communication tower. This matrix form makes the subsequent data processing and optimization prediction process more scientific and flexible, and at the same time provides a clear relationship orientation data set for the construction of the communication tower structure undirected graph, so that the communication tower structure undirected graph can be automatically and quickly constructed according to the connection relationship matrix.

[0113] Through this scheme, by analyzing the connection node position set and the connection component angle set, the number of connection nodes and the connection relationship pointing value can be accurately determined, and a connection relationship matrix for quantifying the relationship between each connection node in the railway communication tower is constructed to construct a communication tower structure undirected graph. The connection relationship matrix is ​​used to provide a clear relationship pointing data set for the construction of the communication tower structure undirected graph, so that the communication tower structure undirected graph can be automatically and quickly constructed according to the connection relationship matrix.

[0114] In some embodiments, according to the communication tower structure undirected graph, the spatial coordinates of each connection node are extracted; based on the spatial coordinates, according to the key installation node depth, the installation node relative distance set and the connection component strength set, the structural feature vector corresponding to each connection node is determined, and the structural feature matrix is ​​constructed based on this, which is specifically the following expression (2):

[0115]

[0116] in, is the structural feature matrix, For the The structural feature vector of the connected nodes, For the The horizontal coordinates of the connected nodes, For the The vertical coordinates of the connected nodes, For the The height coordinates of the connected nodes; For the The key installation node depth of the connection node. If the current connection node is not an installation node, this item is 0; For the The relative distance set of the installation nodes corresponding to the connection nodes. If the current connection node is not an installation node, this item is 0; For the The connection component strength set corresponding to each connection node; based on the connection relationship matrix and the structural feature matrix, the structural aggregation feature vector of each connection node is determined, and the structural static feature vector set is constructed based on this.

[0117] The spatial coordinates may be the three-dimensional spatial position coordinates corresponding to the connection node on the railway communication tower.

[0118] The structural aggregation feature vector can be constructed based on the connection node, aggregating the structural characteristics and static strength performance of the connection node to obtain vector information used to characterize the local static characteristics of the railway communication tower corresponding to different connection nodes.

[0119] Specifically, by extracting the attribute values ​​of each node in the undirected graph of the communication tower structure, the spatial coordinates corresponding to each connection node are determined, providing a spatial calculation basis for the design and analysis of the local structural characteristics of the railway communication tower. In combination with the key installation node depth, installation node relative distance set and connection component strength set obtained in the above-mentioned embodiment, the feature set corresponding to different connection points is constructed through expression (2). By establishing a structural feature matrix, the features of all connection nodes can be clearly collected and displayed, which is convenient for subsequent quantitative analysis, performance simulation and stability prediction. At the same time, the structural feature matrix provides necessary data support for subsequent numerical calculations and machine learning model training, so that the final decision-making process has a scientific basis. Further, through mathematical analysis, according to the connection relationship matrix and the structural feature matrix, a structural aggregation feature vector that can reflect the stability performance of the structural features of the connection nodes under the influence of different connection relationships is constructed.

[0120] Through this scheme, the undirected graph of the communication tower structure is analyzed, and the spatial coordinates corresponding to each connection node are extracted. On this basis, the key installation node depth, the relative distance set of the installation nodes and the connection component strength set corresponding to each connection node are combined, and a structural feature matrix is ​​constructed through mathematical analysis. The characteristics of each connection node are integrated and described using mathematical language to facilitate subsequent quantitative analysis, performance simulation and stability prediction. Further, based on the connection relationship matrix and the structural feature matrix, a structural aggregation feature vector is constructed that can reflect the stability performance of the structural characteristics of the connection node under the influence of different connection relationships.

[0121] In some embodiments, the number of connection paths corresponding to each connection node is determined according to the connection component strength set corresponding to each connection node; based on the number of connection paths, the structural aggregation feature vector of each connection node is determined according to the connection relationship matrix and the structural feature matrix, which is specifically the following formula (3):

[0122]

[0123] in, For the The aggregate feature vector corresponding to the connected nodes, For the The number of connection paths corresponding to the connection nodes, is the number of connected nodes, For the The connection node and The connection relationship between the connection nodes points to the value, For the The structural feature vector of the connected nodes.

[0124] The number of connection paths may be the total number of connection paths formed by connection components between a connection node and other connection nodes having a connection relationship with it. The number of connection paths corresponding to a connection node is the same as the number of connection components existing between the connection node and other connection nodes.

[0125] Specifically, in the railway communication tower structure, the number of connection paths connecting nodes reflects the complexity of the relative positions and interactions of the nodes. The characteristics of different connection nodes are weighted by formula (3) to calculate the aggregate feature vector , comprehensively measuring the state and influence of a connected node in the entire structure, is crucial for evaluating the overall stability of a communication tower, because the tilt of a communication tower may be caused by the instability of some connected nodes after being affected. Understanding the degree of connection of each node in the network and its characteristics can help identify potential risk factors. By introducing aggregated feature vectors, the interaction between the information of connected nodes and their environment can be more comprehensively captured, thereby improving the accuracy of subsequent tilt risk prediction.

[0126] Through this scheme, by means of mathematical analysis, on the basis of the number of connection paths, according to the connection relationship matrix and the structural feature matrix, the corresponding mathematical formula is designed to achieve accurate and scientific quantification of the aggregated feature vectors of the connection nodes. By introducing the aggregated feature vectors, the interaction between the information of the connection nodes and their environment can be more comprehensively captured, thereby improving the accuracy of subsequent tilt risk prediction.

[0127] In some embodiments, based on meteorological change information, wind change information, temperature change information and meteorological carrier change information are extracted; based on geological change information, soil bearing change information, vibration conduction change information and settlement change information are extracted; based on wind change information, temperature change information, meteorological carrier change information, soil bearing change information, vibration conduction change information and settlement change information, a dynamic change feature information set is constructed.

[0128] The environmental factor information set includes meteorological change information and geological change information. The meteorological change information can be; the geological change information can be

[0129] The wind force variation information may be information on the wind speed variation over time in the area where the railway communication tower is located.

[0130] The temperature change information may be information about the temperature change over time in the area where the railway communication tower is located.

[0131] The information on changes in meteorological carriers may be information on changes over time in the excess impurity content in the air caused by meteorological changes in the area where the railway communication tower is located, such as changes in dust content, changes in snow volume, etc.

[0132] The soil bearing change information may be information on the time-varying ability of the soil at the base of the communication tower to support the load applied by the superstructure without causing damage or significant settlement.

[0133] The vibration transmission change information may be information on the time-varying intensity of ground or underground vibrations transmitted to the base of a communication tower.

[0134] The settlement change information may be information on the change over time of the vertical displacement of the soil at the base of the communication tower due to load, natural factors or other environmental factors.

[0135] Specifically, the tilt risk of railway communication towers will be affected by a series of dynamic environmental factors in the area where they are located. The dynamic environmental factors can be divided into meteorological factors and geological factors using two-dimensional division thinking. From the spatial dimension, meteorological factors affect the tilt risk of railway communication towers from top to bottom, while geological factors affect the tilt risk of railway communication towers from bottom to top. Through the fusion analysis of meteorological factors and geological factors, a comprehensive analysis of the impact of environmental factors can be conducted from the spatial dimension; among them, wind change information mainly reflects the wind change in the area where the railway communication tower is located. Strong winds will increase the lateral pressure of the communication tower, which will cause the tower to tilt or deform, especially in high-rise structures or communication towers with lighter materials. The impact of wind is more significant; temperature change information reflects the temperature change in the area where the railway communication tower is located. Temperature changes can cause thermal expansion or contraction of materials. If the structural materials of the communication tower expand at high temperatures and contract at low temperatures, it may cause stress changes in the connection parts, thereby affecting the stability of the entire structure; the meteorological carrier change information reflects the changes in air carriers caused by the weather, which may cause erosion (such as wind and sand erosion) or additional loads (such as snow accumulation) on the railway communication tower; changes in soil bearing capacity may cause the foundation of the communication tower to sink or settle unevenly, directly increasing the risk of tilting; vibration conduction change information can reflect the impact of vibrations caused by geological activities (such as earthquakes, traffic vibrations, etc.) in the area where the railway communication tower is located on the communication tower. If the vibration energy is too strong, it may cause excessive displacement of the tower structure; settlement change information can directly reflect the changes in the foundation force of the railway communication tower; by integrating the above various change information, the dynamic change feature information set constructed can provide scientific data support for the subsequent all-round prediction of the risk of tilting the communication tower.

[0136] Through this scheme, the dynamic environmental factors that affect the risk of railway communication tower tilt are divided into two dimensions according to the spatial dimension, and the wind change information, temperature change information, meteorological carrier change information, soil bearing change information, vibration conduction change information and settlement change information in the meteorological change information and geological change information are extracted respectively. Based on this, a dynamic change feature information set is constructed to provide scientific data support for the subsequent all-round prediction of the risk of communication tower tilt.

[0137] In some embodiments, a preset recurrent neural network model is used to perform time series analysis on each dynamic change feature in the dynamic change feature information to determine a dynamic time series feature vector set; based on the dynamic time series feature vector set, according to the structural static feature vector set, the fusion influence factors of different dynamic features at different time points on the structural feature vector corresponding to each connection node are determined; based on the structural static feature vector set, according to the fusion influence factors, the comprehensive tilt risk at different time points is determined, and a tilt risk prediction model is constructed based on this.

[0138] The preset recurrent neural network model may be a preset mathematical model for performing time series analysis on input data. The preset recurrent neural network model may adopt a basic RNN (Recurrent Neural Network) or a special RNN, such as an LSTM (Long Short-Term Memory Network).

[0139] Time series analysis can be the process of analyzing the data series of dynamic environmental factors that change over time. The purpose is to analyze the trends, cycles and irregular changes of dynamic environmental factors over time.

[0140] The dynamic time series feature vector set may be a feature vector extracted after time series analysis of the dynamic change feature, and is used to represent the state of the dynamic change feature changing over time.

[0141] The fusion impact factor may be a mathematical factor reflecting the degree of influence of dynamic features on node structural features.

[0142] The comprehensive tilt risk can be the result of a comprehensive risk assessment of railway communication towers based on the fusion impact of dynamic and static characteristics, combined with historical data and current dynamic characteristics.

[0143] Specifically, by presetting a recurrent neural network model, a time series analysis is performed on each dynamic change feature in the dynamic change feature information, and a dynamic time series feature vector set is determined, so that the interaction of dynamic features is clearly visible, laying the foundation for subsequent models and ensuring that the tilt risk prediction model can process time-dependent data; by mathematical analysis, a fusion analysis is performed on the dynamic time series feature vector set and the structural static feature vector set to obtain the corresponding fusion influence factor, making full use of the stability and environmental impact of the communication tower itself to accurately assess the risk; further, by using mathematical analysis, according to the structural static feature vector set and the fusion influence factor, the comprehensive tilt risk at different time points is accurately obtained, so as to construct a tilt risk prediction model that can predict the tilt risk of railway communication towers within sufficient time margin.

[0144] Through this scheme, a time series analysis is performed on each dynamic change feature in the dynamic change feature information to determine a dynamic time series feature vector set, so as to capture the dynamic changes of the railway communication tower affected by time and external environmental factors during operation. On the basis of the dynamic time series feature vector set, combined with the structural static feature vector set, a structural fusion influencing factor that can reflect the degree of influence of different dynamic features on the static structure in different time bags is determined, and then the comprehensive tilt risk at different time points is determined, so as to construct a tilt risk prediction model that can predict the tilt risk of the railway communication tower within a sufficient time margin.

[0145] In some embodiments, based on the dynamic time series feature vector set and according to the structural static feature vector set, the fusion influence factor of different dynamic features at different time points on the structural feature vector corresponding to each connection node is determined, specifically, the following formula (4):

[0146]

[0147] in, is the fusion impact factor, is the preset ratio conversion coefficient, For the The structural feature vector of the connected nodes, is the first in the dynamic time series feature vector set Dynamic features at time points The corresponding vector value under For the current The corresponding preset feature influence coefficient.

[0148] The preset proportional conversion coefficient can be a numerical value used to adjust the influence scale of the fusion influence factor. The preset proportional conversion coefficient can adopt a convolutional neural network, input dynamic features, and output the tilt state of the connection node, and then gradually adjust it in the model through the back propagation algorithm.

[0149] The preset feature influence coefficient can be a numerical value used to characterize the degree of influence of dynamic features on structural features. The preset feature influence coefficient can be initially set through expert experience or regression analysis of historical data. During the model training phase, the preset feature influence coefficient is set as a learnable parameter, and its value is adjusted through an optimization algorithm (such as gradient descent) so that the final output fusion influence factor is more in line with the actually observed tilt state.

[0150] Specifically, formula (4) is used to conduct a combined analysis of the impact of different dynamic characteristics on the current connection point at different time points, thereby capturing the comprehensive impact of environmental changes on the connection nodes, and thereby quantifying the fusion impact factor to improve the accuracy of railway communication tower tilt risk prediction.

[0151] Through this scheme, mathematical analysis methods are used to design mathematical formulas based on the dynamic time series feature vector set and the structural static feature vector set. The influence of different dynamic features on the current connection point at different time points is combined and analyzed, and the comprehensive impact of environmental changes on the connection nodes is captured. The fusion impact factor is quantified to improve the accuracy of the prediction of the risk of railway communication tower tilt.

[0152] In some embodiments, based on the structural static feature vector set, the comprehensive tilt risk at different time points is determined according to the fusion impact factor, specifically the following formula (5):

[0153]

[0154] in, For time point The corresponding comprehensive tilt risk is is the preset risk conversion factor, is the number of connected nodes, For the The structural feature vector of the connected nodes, It is the fusion impact factor.

[0155] The preset risk conversion coefficient can be a conversion coefficient used to adjust the mapping relationship between the predicted tilt risk and the actual tilt risk. An initial value can be set for the preset risk conversion coefficient based on previous historical data or industry standards, and through a regression analysis model, the error between the comprehensive tilt risk predicted by the model and the actually observed tilt risk can be minimized to obtain an accurate preset risk conversion coefficient.

[0156] Specifically, the contribution of each connection point to the overall tilt risk of the railway communication tower under the influence of dynamic characteristics is superimposed by formula (5), and then the local characteristic changes of each connection node are summarized into an overall risk assessment to quantify the comprehensive tilt risk that can standardize the overall tilt risk of the railway communication tower. The natural logarithm is introduced in the form of to reflect the nonlinear growth characteristics of the tilt risk, and to avoid excessive linear growth, while ensuring that the risk value will not be negative, thus ensuring the stability of the mathematical model.

[0157] Through this scheme, mathematical analysis methods are used to design mathematical formulas based on the set of structural static feature vectors and fusion influencing factors. The contribution of each connection point to the overall tilt risk of the railway communication tower under the influence of dynamic characteristics is superimposed, and then the local characteristic changes of each connection node are summarized into an overall risk assessment, so as to quantify the comprehensive tilt risk that can standardize the overall tilt risk of the railway communication tower, and significantly improve the prediction accuracy of the tilt risk of the railway communication tower.

[0158] Figure 3 A schematic diagram of the structure of a communication tower risk monitoring system based on deep learning provided in one embodiment of the present application is shown in FIG. Figure 3 As shown, a communication tower risk monitoring system 300 based on deep learning in this embodiment includes: a structure analysis module 301, a dynamic feature analysis module 302, a model building module 303 and a risk output module 304.

[0159] The structural analysis module 301 is used to obtain a structural static parameter set and determine a structural feature information set according to the structural static parameter set;

[0160] A dynamic feature analysis module 302 is used to obtain an environmental factor information set, analyze the environmental factor information set, and determine a dynamic change feature information set;

[0161] A model building module 303 is used to perform fusion optimization analysis on the structural feature information set and the dynamic change feature information set according to the structural feature information set and the dynamic change feature information set to build a tilt risk prediction model;

[0162] The risk output module 304 is used to determine and output tilt risk change information according to the tilt risk prediction model.

[0163] Optionally, the structural analysis module 301 is specifically used for:

[0164] Determine a connection node position set and a connection component angle set according to the communication tower structure scanning information;

[0165] Constructing an undirected graph of communication tower structures according to the connection node position set and the connection component angle set;

[0166] Analyze the communication tower installation information, and extract the key installation node depth and installation node relative distance set corresponding to each installation node;

[0167] Based on the material information of the communication tower, the connection node position set and the connection component angle set are analyzed to determine the connection component strength between any two connected connection nodes, thereby constructing a connection component strength set corresponding to each connection node;

[0168] Based on the communication tower structure undirected graph, constructing a set of structural static feature vectors according to the key installation node depth, the installation node relative distance set and the connection component strength set;

[0169] The structural static feature vector set is used as the structural feature information set.

[0170] Optionally, when constructing the communication tower structure undirected graph according to the connection node position set and the connection component angle set, the structure analysis module 301 is specifically used to:

[0171] Analyze the connection node position set and the connection component angle set to determine the number of connection nodes and the connection relationship pointing value between any two connection nodes;

[0172] According to the number of connection nodes and the connection relationship, a connection relationship matrix is ​​constructed, which is specifically the following expression:

[0173] ;

[0174] in, is the connection relationship matrix, is the number of connected nodes, Representative The connection node and The connection relationship between the connection nodes points to a value;

[0175] Based on the connection relationship matrix, the communication tower structure undirected graph is constructed according to the connection node position set and the connection component angle set.

[0176] Optionally, when the structural analysis module 301 constructs a structural static feature vector set based on the communication tower structure undirected graph, according to the key installation node depth, the installation node relative distance set and the connection component strength set, it is specifically used to:

[0177] Extracting the spatial coordinates of each connection node according to the communication tower structure undirected graph;

[0178] Based on the spatial coordinates, according to the key installation node depth, the installation node relative distance set and the connection component strength set, the structural feature vector corresponding to each connection node is determined, and a structural feature matrix is ​​constructed based on this, which is specifically expressed as follows:

[0179] ;

[0180] in, is the structural feature matrix, For the The structural feature vectors of the connected nodes, For the The horizontal coordinate of the connection node, For the The vertical coordinate of each of the connection nodes, For the The height coordinates of the connection nodes; For the The key installation node depth of the connection node. If the current connection node is not the installation node, this item is 0; For the The relative distance set of the installation nodes corresponding to the connection nodes. If the current connection node is not the installation node, this item is 0; For the A set of connection component strengths corresponding to each of the connection nodes;

[0181] Based on the connection relationship matrix and the structural feature matrix, the structural aggregation feature vector of each connection node is determined, and the structural static feature vector set is constructed based on this.

[0182] Optionally, when the structural analysis module 301 determines the structural aggregation feature vector of each connection node based on the connection relationship matrix and the structural feature matrix, it is specifically used to:

[0183] Determine the number of connection paths corresponding to each connection node according to the connection component strength set corresponding to each connection node;

[0184] Based on the number of connection paths, according to the connection relationship matrix and the structural feature matrix, the structural aggregation feature vector of each connection node is determined, specifically as follows:

[0185] ;

[0186] in, For the The aggregate feature vector corresponding to the connected nodes, For the The number of connection paths corresponding to the connection nodes, is the number of connected nodes, For the The connection node and The connection relationship between the connection nodes points to a value, For the The structural feature vectors of the connection nodes.

[0187] Optionally, the dynamic feature analysis module 302 is specifically used to:

[0188] Extracting wind force change information, temperature change information and meteorological carrier change information according to the meteorological change information;

[0189] Extracting soil bearing change information, vibration conduction change information and settlement change information according to the geological change information;

[0190] The dynamic change characteristic information set is constructed based on the wind force change information, temperature change information, meteorological carrier change information, soil bearing change information, vibration conduction change information and settlement change information.

[0191] Optionally, the model building module 303 is specifically used to:

[0192] By using a preset recurrent neural network model, a time series analysis is performed on each dynamic change feature in the dynamic change feature information to determine a dynamic time series feature vector set;

[0193] Based on the dynamic time series feature vector set and according to the structural static feature vector set, determining the fusion influence factor of different dynamic features at different time points on the structural feature vector corresponding to each connection node;

[0194] Based on the structural static feature vector set and according to the fusion influencing factors, the comprehensive tilt risk at different time points is determined, and the tilt risk prediction model is constructed based on this.

[0195] Optionally, when the model building module 303 determines the fusion impact factor of different dynamic features at different time points on the structural feature vector corresponding to each connection node based on the dynamic time series feature vector set and the structural static feature vector set, it is specifically the following formula:

[0196] ;

[0197] in, is the fusion impact factor, is the preset ratio conversion coefficient, For the The structural feature vectors of the connection nodes, is the first Dynamic features at time points The corresponding vector value under For the current The corresponding preset feature influence coefficient.

[0198] ;

[0199] in, For time point The corresponding comprehensive tilt risk is as follows: is the preset risk conversion factor, is the number of connected nodes, For the The structural feature vectors of the connection nodes, is the fusion influencing factor.

[0200] The system of this embodiment can be used to execute the method of any of the above embodiments. The implementation principles and technical effects are similar and will not be described in detail here.

Claims

1. A communication tower risk monitoring method based on deep learning, characterized in that: include: Acquire a structural static parameter set, and determine a structural feature information set according to the structural static parameter set; Acquire an environmental factor information set, analyze the environmental factor information set, and determine a dynamic change characteristic information set; According to the structural feature information set and the dynamic change feature information set, a fusion analysis is performed on the structural feature information set and the dynamic change feature information set to construct a tilt risk prediction model; Determining and outputting tilt risk change information according to the tilt risk prediction model; The structural static parameter set includes communication tower structure scanning information, communication tower installation information and communication tower material information. The structural characteristic information set is determined according to the structural static parameter set, including: Determine a connection node position set and a connection component angle set according to the communication tower structure scanning information; Constructing an undirected graph of communication tower structures according to the connection node position set and the connection component angle set; Analyze the communication tower installation information, and extract the key installation node depth and the installation node relative distance set corresponding to each installation node; the key installation node depth is used to characterize the depth of the installation node in the medium with which the installation node is hard-connected; the medium includes soil; Based on the material information of the communication tower, the connection node position set and the connection component angle set are analyzed to determine the connection component strength between any two connected connection nodes, thereby constructing a connection component strength set corresponding to each connection node; Based on the communication tower structure undirected graph, constructing a set of structural static feature vectors according to the key installation node depth, the installation node relative distance set and the connection component strength set; The structural static feature vector set is used as the structural feature information set.

2. The method according to claim 1, characterized in that: The step of constructing a communication tower structure undirected graph according to the connection node position set and the connection component angle set includes: Analyze the connection node position set and the connection component angle set to determine the number of connection nodes and the connection relationship pointing value between any two connection nodes; the connection relationship pointing value is a numerical value used to characterize whether there is a connection relationship between the two connection nodes, and the connection relationship pointing value is set as follows: when there is a connection relationship between the two connection nodes, the corresponding connection relationship pointing value is 1, and when there is no connection relationship between the two connection nodes, the corresponding connection relationship pointing value is 0; According to the number of connection nodes and the connection relationship, a connection relationship matrix is ​​constructed, which is specifically the following expression: ； in, is the connection relationship matrix, is the number of connected nodes, Representative The connection node and The connection relationship between the connection nodes points to a value; Based on the connection relationship matrix, the communication tower structure undirected graph is constructed according to the connection node position set and the connection component angle set.

3. The method according to claim 2, characterized in that: The method of constructing a structural static feature vector set based on the communication tower structure undirected graph according to the key installation node depth, the installation node relative distance set and the connection component strength set includes: Extracting the spatial coordinates of each connection node according to the communication tower structure undirected graph; Based on the spatial coordinates, according to the key installation node depth, the installation node relative distance set and the connection component strength set, the structural feature vector corresponding to each connection node is determined, and the structural feature matrix is ​​constructed based on this, which is specifically expressed as follows: ； in, is the structural feature matrix, For the The structural feature vectors of the connected nodes, For the The horizontal coordinate of the connection node, For the The vertical coordinate of the connection node, For the The height coordinates of the connection nodes; For the The key installation node depth of the connection node. If the current connection node is not the installation node, this item is 0; For the The relative distance set of the installation nodes corresponding to the connection nodes. If the current connection node is not the installation node, this item is 0; For the A set of connection component strengths corresponding to each of the connection nodes; Based on the connection relationship matrix and the structural feature matrix, the structural aggregation feature vector of each connection node is determined, and the structural static feature vector set is constructed based on this.

4. The method according to claim 3, characterized in that: The determining, based on the connection relationship matrix and according to the structural feature matrix, a structural aggregation feature vector of each connection node comprises: Determine the number of connection paths corresponding to each connection node according to the connection component strength set corresponding to each connection node; Based on the number of connection paths, according to the connection relationship matrix and the structural feature matrix, the structural aggregation feature vector of each connection node is determined, specifically as follows: ； in, For the The aggregate feature vector corresponding to the connected nodes, For the The number of connection paths corresponding to the connection nodes, is the number of connected nodes, For the The connection node and The connection relationship between the connection nodes points to a value, For the The structural feature vectors of the connection nodes.

5. The method according to claim 4, characterized in that The environmental factor information set includes meteorological change information and geological change information. The step of analyzing the environmental factor information set to determine the dynamic change characteristic information set includes: Extracting wind force change information, temperature change information and meteorological carrier change information according to the meteorological change information; Extracting soil bearing change information, vibration conduction change information and settlement change information according to the geological change information; The dynamic change characteristic information set is constructed based on the wind force change information, temperature change information, meteorological carrier change information, soil bearing change information, vibration conduction change information and settlement change information.

6. The method according to claim 5, characterized in that The step of performing fusion analysis on the structural feature information set and the dynamic change feature information set according to the structural feature information set and the dynamic change feature information set to construct a tilt risk prediction model includes: By using a preset recurrent neural network model, a time series analysis is performed on each dynamic change feature in the dynamic change feature information to determine a dynamic time series feature vector set; Based on the dynamic time series feature vector set and according to the structural static feature vector set, determining the fusion influence factor of different dynamic features at different time points on the structural feature vector corresponding to each connection node; Based on the structural static feature vector set and according to the fusion influencing factors, the comprehensive tilt risk at different time points is determined, and the tilt risk prediction model is constructed based on this.

7. The method according to claim 6, characterized in that Based on the dynamic time series feature vector set, according to the structural static feature vector set, the fusion influence factor of different dynamic features at different time points on the structural feature vector corresponding to each connection node is determined, which is specifically the following formula: ； in, is the fusion impact factor, is the preset ratio conversion coefficient, For the The structural feature vectors of the connection nodes, is the first Dynamic features at time points The corresponding vector value under For the current The corresponding preset feature influence coefficient.

8. The method according to claim 7, characterized in that Based on the structural static feature vector set, the comprehensive tilt risk at different time points is determined according to the fusion impact factor, which is specifically the following formula: ； in, For time point The corresponding comprehensive tilt risk is as follows: is the preset risk conversion factor, is the number of connected nodes, For the The structural feature vectors of the connection nodes, is the fusion influencing factor.

9. A communication tower risk monitoring system based on deep learning, characterized in that: The method as claimed in any one of claims 1 to 8 comprises: A structural analysis module, used to obtain a structural static parameter set, and determine a structural feature information set according to the structural static parameter set; A dynamic feature analysis module, used to obtain an environmental factor information set, analyze the environmental factor information set, and determine a dynamic change feature information set; A model building module, used to perform fusion optimization analysis on the structural feature information set and the dynamic change feature information set according to the structural feature information set and the dynamic change feature information set, so as to build a tilt risk prediction model; The risk output module is used to determine and output tilt risk change information according to the tilt risk prediction model.

Images