Early warning method and system based on tunnel historical event feature analysis

Abstract

The invention relates to the technical field of tunnel early warning, in particular to an early warning method and system based on tunnel historical event feature analysis, and the method comprises the following steps: S1, historical event data collection and preprocessing: collecting historical event data in a tunnel; s2, historical event classification and induction: automatically classifying historical event data; s3, based on the event association network and data analysis, establishing the event association network of tunnel historical events; s4, event feature extraction and model establishment: key features are extracted from historical event data, and a historical event feature model is established; and S5, real-time monitoring data fusion and analysis: realizing comprehensive monitoring and analysis of the internal state of the tunnel. According to the invention, through deep data insight, the early warning system can make a response based on scientific analysis, so that the data processing efficiency and the prediction accuracy are remarkably improved.

Description

Technical Field

[0001] The present invention relates to the technical field of tunnel early warning, and in particular to an early warning method and system based on characteristic analysis of tunnel historical events. Background Art

[0002] In the field of tunnel management and monitoring, tunnel safety has always been an important research and application direction. With the frequent occurrence of tunnel accidents, such as geological disasters, traffic accidents and various accidents caused by environmental factors, there is an urgent need for a system that can efficiently predict and manage these events. Traditional tunnel monitoring methods mostly rely on manual monitoring or simple sensor data collection. These methods are often difficult to analyze data in real time and accurately, especially in processing complex data and predicting potential risks.

[0003] In existing technologies, although some advanced monitoring systems have begun to use automation technology to collect data, these systems usually lack the ability to deeply analyze historical data and combine it with real-time data, and cannot effectively predict and prevent potential risks. In addition, the accuracy of data processing and event warnings of these systems needs to be improved, and they often fail to accurately understand the correlation between events and the impact transmission path, resulting in insufficient or excessive responses. Summary of the invention

[0004] Based on the above objectives, the present invention provides an early warning method and system based on characteristic analysis of tunnel historical events.

[0005] An early warning method based on characteristic analysis of tunnel historical events comprises the following steps:

[0006] S1, historical event data collection and preprocessing: collect historical event data inside the tunnel, and clean and standardize the collected historical event data;

[0007] S2, historical event classification and induction: automatically classify historical event data and establish an event type database;

[0008] S3, based on event association network and data analysis: based on the results of historical event classification and induction, an event association network of tunnel historical events is established to reveal the implicit relationship between events and the impact transmission path;

[0009] S4, event feature extraction and model building: based on the analysis results of the event association network, key features are extracted from the historical event data. At the same time, the historical event feature model is established by combining the key features of the historical event data, comprehensively considering the impact of various factors on tunnel safety;

[0010] S5, real-time monitoring data fusion and analysis: The real-time monitoring data is integrated with the historical event feature model to achieve comprehensive monitoring and analysis of the internal status of the tunnel, timely detect abnormal situations, and through anomaly detection algorithms and early warning rules, according to the degree of matching between the real-time monitoring data and the historical event feature model, determine whether there is a risk inside the tunnel and trigger the corresponding early warning mechanism.

[0011] Furthermore, the historical event data collection and preprocessing in S1 includes:

[0012] S11, historical event data collection: a variety of sensors and monitoring equipment are arranged inside the tunnel, including earthquake sensors, temperature and humidity sensors, water level sensors and water quality monitoring equipment, to collect various historical event data, including geological disasters, traffic accidents, earthquake activities, temperature and humidity changes, and water level changes;

[0013] S12, data preprocessing: clean the collected historical event data, remove duplicate data and abnormal data, and unify the format and unit of the cleaned historical event data.

[0014] Furthermore, the classification and summarization of historical events in S2 includes:

[0015] S21, historical event data classification: automatically classify historical event data and classify the historical event data into corresponding categories according to the characteristics of the event;

[0016] S22, establishing an event type database: according to the classification results, establishing a historical event type database, and recording relevant information of different types of historical event data, the relevant information including event characteristics, occurrence time and location.

[0017] Furthermore, the event-related network and data analysis in S3 includes:

[0018] S31, establish event association network: based on the results of historical event classification and induction, use graph theory to construct the event association network of tunnel historical events;

[0019] S32, Analyze the event correlation network: Analyze the event correlation network to reveal the implicit relationship and impact transmission path between historical events. By exploring the correlation and impact degree between historical events, we can understand the development law of events inside the tunnel.

[0020] Furthermore, the graph theory method in S31 specifically includes:

[0021] S311, determining nodes and edges: representing each historical event as a node in the network, each node representing a type of historical event, and determining associations between historical events, and representing the associations as edges in the network;

[0022] S312, defining an adjacency matrix and determining a threshold: the association relationship between nodes is represented as an adjacency matrix, and based on the empirical threshold (the experience of domain experts or the results of previous studies), statistical distribution (the distribution of event association relationships), network density (the density of the event association network), and visual analysis (the topological structure of the network and the weight distribution of the edges), a threshold is determined to screen the association relationship of historical events;

[0023] S313, constructing an event association network: according to the adjacency matrix and the threshold, an event association network is established. For association relationships greater than the threshold, they are represented as edges in the event association network, and association relationships less than the threshold are ignored.

[0024] Furthermore, the analysis event correlation network in S32 includes:

[0025] S321, Node degree centrality analysis: Node degree centrality measures the number of connections of each node in the event association network, that is, the importance of the node. The calculation formula is:

[0026]

[0027] Among them, k i represents the number of connections of node i, and n represents the total number of nodes in the network;

[0028] S322, Node betweenness centrality analysis: Node betweenness centrality measures the degree of intermediary of each node in the event association network, that is, the importance of the node in the event propagation path. The calculation formula is:

[0029]

[0030] Among them, σ st represents the number of shortest paths from node s to node t, σ st (i) represents the number of nodes passed through in the shortest path from node s to node t;

[0031] S323, community discovery analysis: Community discovery aggregates the connected nodes in the event association network into communities, reveals the implicit relationship between events and the impact transmission path, and uses the Louvain algorithm to optimize the modularity index of the network to achieve community discovery. The calculation formula is:

[0032]

[0033] Among them, A ijrepresents the connection strength between node i and node j, k i and k j They represent the degrees of node i and node j respectively, m represents the total connection strength of the network, and c i and c j denote the communities to which nodes i and j belong respectively, δ(c i ,c j ) is the Kroneckerdelta function, when c i =c j The value is 1 when , otherwise it is 0.

[0034] Furthermore, the event feature extraction and model building in S4 includes:

[0035] S41, extract key features: based on the analysis results of the event association network, extract key features from the historical event data. The key features include the environmental conditions (such as temperature and humidity), geographical location (latitude and longitude coordinates), duration and impact range (affected area or range), traffic flow, traffic operation status and lane closure at the time of the event;

[0036] S42, establishing a historical event feature model: establishing a historical event feature model by combining the extracted key features with relevant information of the historical event data, and training the historical event feature model based on the historical event data to predict the probability and impact of the event;

[0037] S43, model optimization: comprehensively consider the impact of various factors on tunnel safety, including geological disasters, traffic accidents, seismic activities, climate conditions and geographical location, and optimize the parameters of the historical event characteristic model.

[0038] Furthermore, the historical event feature model adopts a Bayesian network model, and the Bayesian network model includes:

[0039] Define the nodes of the Bayesian network: Each node represents a historical event type, including geological disasters, traffic accidents, earthquake activities, temperature changes, and water level changes. Each historical event node contains multiple attributes related to the historical event, including time, location, and severity;

[0040] Determine the dependencies between nodes: Define the dependencies between nodes based on historical event data and domain knowledge, and use the dependencies to build a network structure, where edges represent conditional dependencies between historical events;

[0041] Parameter learning and Bayesian network construction: In a Bayesian network, the behavior of each node depends on its parent node. The dependency is expressed through the conditional probability table (CPT). i Its parent node set Pa(Xi ), conditional probability P(X i |Pa(X i )) indicates that node X is in the state of the parent node when it is known i The state probability of node X i The state probability of is estimated by the statistical information in the data set, and the calculation formula is:

[0042]

[0043] Among them, N(x i ,pa(x i )) indicates that node X in the data set i and its parent node set pa(x i ) is the number of observations that are in the same state at the same time, N(pa(x i )) is the parent node set pa(x i ) the number of observations in that state;

[0044] Network reasoning and prediction: The reasoning of Bayesian networks uses the belief propagation algorithm to calculate the edge probability of any node in the network. The calculation formula is:

[0045]

[0046] Among them, Bel(X i ) is node X i The belief is from node X j To Node X i The message, α is the normalization constant;

[0047] From Node X j To Node X i The calculation formula for the message is:

[0048]

[0049] Among them, P(X j ∣X i ,Pa(X j )) indicates that when X is known i and X j Other parent nodes Pa(X j ), X j The conditional probability of .

[0050] Furthermore, the real-time monitoring data fusion and analysis in S5 includes:

[0051] S51, data fusion: real-time monitoring data (such as geological sensor data, traffic flow sensor data, meteorological conditions) are integrated with historical event feature models to form a comprehensive data view, and the real-time monitoring data is standardized and normalized;

[0052] S52, anomaly detection and threshold setting: Calculate the anomaly detection score and set the threshold of the anomaly detection score based on experience or statistical analysis. When the anomaly detection score of the real-time monitoring data exceeds the set threshold, it is considered a risk event. The calculation formula of the anomaly detection score is:

[0053]

[0054] Among them, x is the data point monitored in real time, μ and σ are the mean and standard deviation calculated based on historical event data;

[0055] S53, early warning rules: formulate early warning rules based on historical data and expert experience;

[0056] S54, risk assessment and response: Combine real-time monitoring data and historical event feature models, use a linear regression model to conduct event risk assessment, and automatically initiate corresponding early warning mechanisms based on the results of the risk assessment, including activating traffic control signals and issuing public safety notices. The linear regression model is expressed as:

[0057] y=β 0 +β 1 x 1 +β 2 x 2 +…+β n x n +∈;

[0058] Among them, y is the response variable, x 1 ,x 2 ,…,x n is a predictor variable of various factors that affect the occurrence of an event (e.g., geographic location, environmental conditions, etc.), β 0 ,β 1 ,…,β n are model parameters and ∈ is the error term.

[0059] An early warning system based on tunnel historical event feature analysis is used to implement the above-mentioned early warning method based on tunnel historical event feature analysis, and includes the following modules:

[0060] Historical event data collection and preprocessing module: responsible for collecting historical event data inside the tunnel, and cleaning and standardizing the historical event data;

[0061] Historical event classification and summarization module: automatically classifies the collected historical event data and establishes an event type database;

[0062] Event association network and data analysis module: Based on the classification results of the historical event classification and induction module, an event association network of tunnel historical events is established to analyze and reveal the implicit relationship and impact transmission path between events;

[0063] Event feature extraction and model building module: extract key features from historical event data based on the analysis results of the event association network, combine the extracted key features to build a historical event feature model, and comprehensively consider the impact of various factors on tunnel safety;

[0064] Real-time monitoring data fusion and analysis module: The real-time monitoring data is integrated with the established historical event feature model to achieve comprehensive monitoring and analysis of the internal status of the tunnel. Through anomaly detection algorithms and early warning rules, abnormal situations are discovered, and based on the degree of match between the real-time monitoring data and the historical event feature model, it is determined whether there is a risk inside the tunnel and the corresponding early warning mechanism is triggered.

[0065] Beneficial effects of the present invention:

[0066] The present invention ensures the quality and consistency of data by systematically collecting and preprocessing historical event data in the tunnel. By utilizing automatic classification and event association network analysis, the system not only reduces the errors and costs of manual operations, but also reveals the implicit relationships and impact paths between events by constructing an accurate event association network. This in-depth data insight enables the early warning system to respond based on scientific analysis, thereby significantly improving the efficiency of data processing and the accuracy of predictions.

[0067] The present invention integrates real-time monitoring data with historical event feature models, and the early warning system can achieve comprehensive monitoring of the internal status of the tunnel. The comprehensive use of real-time data can not only detect abnormal situations in a timely manner, but also quickly determine the risk level inside the tunnel through accurate anomaly detection algorithms and threshold settings. In addition, the system's early warning rules are automatically triggered according to the degree of matching between the actual monitoring data and the historical model, ensuring a rapid and effective response when potential risks are discovered.

[0068] The present invention enhances the systematicness and foresight of tunnel safety management through the establishment of comprehensive analysis and early warning mechanisms. The method enables tunnel operators to take preventive measures before potential safety risks occur, thereby significantly reducing the possibility of accidents and ensuring the safety and stability of tunnel use. This innovative early warning system provides a scientific and efficient solution for tunnel safety management and is a major advancement in the field of tunnel safety technology. BRIEF DESCRIPTION OF THE DRAWINGS

[0069] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0070] Figure 1 A schematic diagram of a warning method flow chart of an embodiment of the present invention;

[0071] Figure 2 Schematic diagram of system function modules according to an embodiment of the present invention. DETAILED DESCRIPTION

[0072] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0073] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The words "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0074] like Figure 1 As shown, an early warning method based on characteristic analysis of tunnel historical events includes the following steps:

[0075] S1, historical event data collection and preprocessing: collect historical event data inside the tunnel, clean and standardize the collected historical event data to ensure data quality and consistency, and provide a reliable data basis for subsequent analysis;

[0076] S2, historical event classification and induction: automatically classify historical event data, establish an event type database, and reduce manual classification costs and error rates;

[0077] S3, based on event association network and data analysis: based on the results of historical event classification and induction, an event association network of tunnel historical events is established to reveal the implicit relationship and impact transmission path between events, providing a basis for further data analysis;

[0078] S4, event feature extraction and model building: based on the analysis results of the event association network, key features are extracted from the historical event data. At the same time, the key features of the historical event data are combined to establish a historical event feature model, comprehensively considering the impact of various factors on tunnel safety, and improving the accuracy and reliability of the early warning model;

[0079] S5, real-time monitoring data fusion and analysis: the real-time monitoring data is integrated with the historical event feature model to achieve comprehensive monitoring and analysis of the internal state of the tunnel, and abnormal situations are discovered in a timely manner. Through the anomaly detection algorithm and early warning rules, according to the matching degree between the real-time monitoring data and the historical event feature model, it is determined whether there is a risk inside the tunnel and the corresponding early warning mechanism is triggered;

[0080] Through the above steps, comprehensive monitoring and early warning of the tunnel safety status are achieved. This method can not only accurately identify the key features of historical events, but also timely discover abnormal situations and take corresponding measures through the fusion analysis of real-time monitoring data, thereby improving the efficiency and reliability of tunnel safety management, reducing accident risks, and ensuring the safety and stability of tunnel use.

[0081] The historical event data collection and preprocessing in S1 include:

[0082] S11, historical event data collection: a variety of sensors and monitoring equipment are arranged inside the tunnel, including earthquake sensors, temperature and humidity sensors, water level sensors and water quality monitoring equipment, to collect various historical event data, including geological disasters, traffic accidents, earthquake activities, temperature and humidity changes, and water level changes;

[0083] S12, data preprocessing: clean the collected historical event data, remove duplicate data and abnormal data, ensure the reliability of data quality, and unify the format and unit of the cleaned historical event data to ensure data consistency and provide a reliable data basis for subsequent analysis.

[0084] The classification and summary of historical events in S2 include:

[0085] S21, historical event data classification: automatically classify historical event data and classify the historical event data into corresponding categories according to the characteristics of the event;

[0086] S22, establish an event type database: according to the classification results, establish a historical event type database, and record relevant information of different types of historical event data, including event characteristics, occurrence time and location, to provide data support for subsequent analysis.

[0087] Event-related network and data analysis in S3 include:

[0088] S31, establish event association network: based on the results of historical event classification and induction, use graph theory to construct the event association network of tunnel historical events;

[0089] S32, Analyze the event correlation network: Analyze the event correlation network to reveal the implicit relationship and impact transmission path between historical events. By exploring the correlation and impact degree between historical events, we can understand the development law of events inside the tunnel.

[0090] Through the above steps, the event correlation network provides a basis and important information support for further data analysis, helping to more accurately predict changes in the tunnel safety status and take corresponding preventive measures to improve the efficiency and reliability of tunnel safety management.

[0091] The graph theory methods in S31 specifically include:

[0092] S311, determining nodes and edges: representing each historical event as a node in the network, each node representing a type of historical event, and determining associations between historical events, and representing the associations as edges in the network. The edges can be determined based on co-occurrence relationships, similarities, or other association indicators between historical events;

[0093] S312, defining an adjacency matrix and determining a threshold: the association relationship between nodes is represented as an adjacency matrix, and based on the empirical threshold (the experience of domain experts or the results of previous studies), statistical distribution (the distribution of event association relationships), network density (the density of the event association network), and visual analysis (the topological structure of the network and the weight distribution of the edges), a threshold is determined to screen the association relationship of historical events;

[0094] S313, constructing an event association network: according to the adjacency matrix and the threshold, an event association network is established. For association relationships greater than the threshold, they are represented as edges in the event association network, and association relationships less than the threshold are ignored.

[0095] The analysis event correlation network in S32 includes:

[0096] S321, Node degree centrality analysis: Node degree centrality measures the number of connections of each node in the event association network, that is, the importance of the node. The calculation formula is:

[0097]

[0098] Among them, k i represents the number of connections of node i, and n represents the total number of nodes in the network. Through the node degree centrality analysis, we can identify the event nodes with important influence in the event association network, so as to understand the degree of association and influence between historical events;

[0099] S322, Node betweenness centrality analysis: Node betweenness centrality measures the degree of intermediary of each node in the event association network, that is, the importance of the node in the event propagation path. The calculation formula is:

[0100]

[0101] Among them, σ st represents the number of shortest paths from node s to node t, σ st (i) represents the number of nodes in the shortest path from node s to node t. Through the analysis of node betweenness centrality, we can identify event nodes that play a key intermediary role in the event association network, thereby understanding the propagation path and impact transmission mode between historical events;

[0102] S323, community discovery analysis: Community discovery aggregates the connected nodes in the event association network into communities, reveals the implicit relationship between events and the impact transmission path, and uses the Louvain algorithm to optimize the modularity index of the network to achieve community discovery. The calculation formula is:

[0103]

[0104] Among them, A ij represents the connection strength between node i and node j, k i and k j They represent the degrees of node i and node j respectively, m represents the total connection strength of the network, and c i and c j denote the communities to which nodes i and j belong respectively, δ(c i ,c j ) is the Kroneckerdelta function, when c i =c j The value is 1 when it is, otherwise it is 0;

[0105] Through the above analysis method, the implicit relationship and impact transmission path between historical events can be fully revealed. By exploring the correlation and impact degree between events, it is helpful to have a deeper understanding of the development law of events inside the tunnel and provide a scientific basis for the establishment of an early warning system.

[0106] Event feature extraction and model building in S4 include:

[0107] S41, extract key features: based on the analysis results of the event association network, extract key features from the historical event data. The key features include the environmental conditions (such as temperature and humidity), geographical location (latitude and longitude coordinates), duration and impact range (affected area or range), traffic flow, traffic operation status and lane closure at the time of the event;

[0108] S42, establishing a historical event feature model: establishing a historical event feature model by combining the extracted key features with relevant information of the historical event data, and training the historical event feature model based on the historical event data to predict the probability and impact of the event;

[0109] S43, model optimization: comprehensively consider the impact of various factors on tunnel safety, including geological disasters, traffic accidents, seismic activities, climate conditions and geographical location, optimize the parameters of historical event characteristic models, and improve prediction accuracy and reliability;

[0110] Through the above content, key features can be accurately extracted from historical event data and an efficient historical event feature model can be established. This model comprehensively considers the impact of various factors on tunnel safety, improves the accuracy and reliability of the early warning system, and provides important support for tunnel safety management. It helps to discover hidden dangers in a timely manner and take effective measures to ensure the safe operation of tunnels and related facilities.

[0111] The historical event feature model adopts the Bayesian network model, which includes:

[0112] Define the nodes of the Bayesian network: Each node represents a historical event type, including geological disasters, traffic accidents, earthquake activities, temperature changes, and water level changes. Each historical event node contains multiple attributes related to the historical event, including time, location, and severity;

[0113] Determine dependencies between nodes: Define dependencies between nodes based on historical event data and domain knowledge. For example, geological disasters may depend on recent earthquake activity and rainfall. Use dependencies to build a network structure, where edges represent conditional dependencies between historical events.

[0114] Parameter learning and Bayesian network construction: In a Bayesian network, the behavior of each node depends on its parent node. The dependency is expressed through the conditional probability table (CPT). i Its parent node set Pa(X i ), conditional probability P(X i |Pa(X i )) indicates that node X is in the state of the parent node when it is known iThe state probability of node X i The state probability of is estimated by the statistical information in the data set, and the calculation formula is:

[0115]

[0116] Among them, N(x i ,pa(x i )) indicates that node X in the data set i and its parent node set pa(x i ) is in the same state at the same time, that is, when the parent node set pa(x i ) is a certain combination of values ​​(for example, rainfall is "high" and seismic activity is "significant"), node X i The number of times a specific value (such as "geological disaster" is "high") appears, N(pa(x i )) is the parent node set pa(x i ) the number of observations in that state;

[0117] Network reasoning and prediction: The reasoning of Bayesian networks uses the belief propagation algorithm to calculate the edge probability of any node in the network. The calculation formula is:

[0118]

[0119] Among them, Bel(X i ) is node X i The belief is from node X j To Node X i , α is a normalization constant to ensure that the sum of probabilities is 1;

[0120] From Node X j To Node X i The calculation formula for the message is:

[0121]

[0122] Among them, P(X j ∣X i ,Pa(X j )) indicates that when X is known i and X j Other parent nodes Pa(X j ), X j The conditional probability of

[0123] In the present invention, the Bayesian network model effectively integrates and analyzes various historical event data related to tunnel safety, such as geological disasters, traffic accidents, and environmental changes. By accurately defining the conditional dependencies between events and quantifying these relationships using conditional probability tables (CPTs), the model can capture the complex interactions between events and provide a comprehensive perspective to predict and evaluate the potential impact of these events on tunnel safety. By using algorithms such as belief propagation for network reasoning, the model can not only predict the probability of occurrence of a single event, but also evaluate the impact of multiple events occurring simultaneously, thereby providing a scientific basis for tunnel safety management and the formulation of emergency response strategies, and enhancing the accuracy of prediction and the efficiency of decision-making.

[0124] Real-time monitoring data fusion and analysis in S5 includes:

[0125] S51, Data Fusion: Real-time monitoring data (such as geological sensor data, traffic flow sensor data, meteorological conditions) are integrated with historical event feature models to form a comprehensive data view, enhance the input quality of the prediction model, and standardize and normalize the real-time monitoring data to ensure that the data format is compatible with the historical data model;

[0126] S52, anomaly detection and threshold setting: Calculate the anomaly detection score and set the threshold of the anomaly detection score based on experience or statistical analysis. When the anomaly detection score of the real-time monitoring data exceeds the set threshold, it is considered a risk event. The calculation formula of the anomaly detection score is:

[0127]

[0128] Among them, x is the data point monitored in real time, μ and σ are the mean and standard deviation calculated based on historical event data;

[0129] S53, warning rules: Based on historical data and expert experience, warning rules are formulated. For example, if a geological sensor continuously detects vibration values ​​exceeding a predetermined threshold, or if traffic flow suddenly drops to an abnormal level, the system will trigger a warning;

[0130] S54, Risk Assessment and Response: Combine real-time monitoring data and historical event feature models, use a linear regression model to conduct event risk assessment, and automatically initiate corresponding early warning mechanisms based on the results of the risk assessment, including activating traffic control signals and issuing public safety notices. The linear regression model is expressed as:

[0131] y=β 0 +β 1 x 1 +β 2 x 2 +…+β n x n+∈;

[0132] Among them, y is the response variable, which represents the risk score, and x 1 ,x 2 ,…,x n is a predictor variable of various factors that affect the occurrence of an event (e.g., geographic location, environmental conditions, etc.), β 0 ,β 1 ,…,β n are model parameters, representing the intercept and the influence weight of each predictor variable, ∈ is the error term, representing the part of the variation that the model fails to explain;

[0133] Through the above content, the real-time monitoring data is integrated with the historical event feature model, which effectively realizes the comprehensive monitoring and analysis of the internal status of the tunnel. By using anomaly detection algorithms and early warning rules, the system can promptly detect and respond to potential risk events, greatly improving the safety management capabilities of the tunnel. This method of integrating real-time data and historical knowledge not only improves the accuracy of early warnings, but also enhances the real-time response capabilities of tunnel management, thereby ensuring the safety of traffic participants and the stable operation of the tunnel.

[0134] like Figure 2 As shown, an early warning system based on tunnel historical event feature analysis is used to implement the above-mentioned early warning method based on tunnel historical event feature analysis, including the following modules:

[0135] Historical event data collection and preprocessing module: responsible for collecting historical event data inside the tunnel, and cleaning and standardizing the historical event data;

[0136] Historical event classification and summarization module: automatically classifies the collected historical event data and establishes an event type database;

[0137] Event association network and data analysis module: Based on the classification results of the historical event classification and induction module, an event association network of tunnel historical events is established to analyze and reveal the implicit relationship and impact transmission path between events;

[0138] Event feature extraction and model building module: extract key features from historical event data based on the analysis results of the event association network, combine the extracted key features to build a historical event feature model, and comprehensively consider the impact of various factors on tunnel safety;

[0139] Real-time monitoring data fusion and analysis module: The real-time monitoring data is integrated with the established historical event feature model to achieve comprehensive monitoring and analysis of the internal status of the tunnel. Through anomaly detection algorithms and early warning rules, abnormal situations are discovered, and based on the degree of match between the real-time monitoring data and the historical event feature model, it is determined whether there is a risk inside the tunnel and the corresponding early warning mechanism is triggered.

[0140] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0141] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. An early warning method based on characteristic analysis of tunnel historical events, characterized in that: The following steps are involved: S1, historical event data collection and preprocessing: collect historical event data inside the tunnel, and clean and standardize the collected historical event data; S2, historical event classification and induction: automatically classify historical event data and establish an event type database; S3, based on event association network and data analysis: based on the results of historical event classification and induction, an event association network of tunnel historical events is established to reveal the implicit relationship between events and the impact transmission path; S4, event feature extraction and model building: based on the analysis results of the event association network, key features are extracted from the historical event data. At the same time, the historical event feature model is established by combining the key features of the historical event data, comprehensively considering the impact of various factors on tunnel safety; S5, real-time monitoring data fusion and analysis: The real-time monitoring data is integrated with the historical event feature model to achieve comprehensive monitoring and analysis of the internal status of the tunnel, timely detect abnormal situations, and through anomaly detection algorithms and early warning rules, according to the degree of matching between the real-time monitoring data and the historical event feature model, determine whether there is a risk inside the tunnel and trigger the corresponding early warning mechanism.

2. The early warning method based on tunnel historical event feature analysis according to claim 1 is characterized in that: The historical event data collection and preprocessing in S1 includes: S11, historical event data collection: a variety of sensors and monitoring equipment are arranged inside the tunnel, including earthquake sensors, temperature and humidity sensors, water level sensors and water quality monitoring equipment, to collect various historical event data, including geological disasters, traffic accidents, earthquake activities, temperature and humidity changes, and water level changes; S12, data preprocessing: clean the collected historical event data, remove duplicate data and abnormal data, and unify the format and unit of the cleaned historical event data.

3. The early warning method based on tunnel historical event feature analysis according to claim 1 is characterized in that: The classification and summary of historical events in S2 include: S21, historical event data classification: automatically classify historical event data and classify the historical event data into corresponding categories according to the characteristics of the event; S22, establishing an event type database: according to the classification results, establishing a historical event type database, and recording relevant information of different types of historical event data, the relevant information including event characteristics, occurrence time and location.

4. The early warning method based on tunnel historical event feature analysis according to claim 1 is characterized in that: The event-related network and data analysis in S3 includes: S31, establish event association network: based on the results of historical event classification and induction, use graph theory to construct the event association network of tunnel historical events; S32, Analyze the event correlation network: Analyze the event correlation network to reveal the implicit relationship and impact transmission path between historical events. By exploring the correlation and impact degree between historical events, we can understand the development law of events inside the tunnel.

5. The early warning method based on tunnel historical event feature analysis according to claim 4 is characterized in that: The graph theory method in S31 specifically includes: S311, determining nodes and edges: representing each historical event as a node in the network, each node representing a type of historical event, and determining associations between historical events, and representing the associations as edges in the network; S312, defining an adjacency matrix and determining a threshold: representing the association relationship between nodes as an adjacency matrix, and determining a threshold based on empirical thresholds, statistical distribution, network density, and visual analysis to screen the association relationship of historical events; S313, constructing an event association network: according to the adjacency matrix and the threshold, an event association network is established. For association relationships greater than the threshold, they are represented as edges in the event association network, and association relationships less than the threshold are ignored.

6. The early warning method based on tunnel historical event feature analysis according to claim 5 is characterized in that: The analysis event correlation network in S32 includes: S321, Node degree centrality analysis: Node degree centrality measures the number of connections of each node in the event association network, that is, the importance of the node. The calculation formula is: Among them, k i represents the number of connections of node i, and n represents the total number of nodes in the network; S322, Node betweenness centrality analysis: Node betweenness centrality measures the degree of intermediary of each node in the event association network, that is, the importance of the node in the event propagation path. The calculation formula is: Among them, σ st represents the number of shortest paths from node s to node t, σ st (i) represents the number of nodes passed through in the shortest path from node s to node t; S323, community discovery analysis: Community discovery aggregates the connected nodes in the event association network into communities, reveals the implicit relationship between events and the impact transmission path, and uses the Louvain algorithm to optimize the modularity index of the network to achieve community discovery. The calculation formula is: Among them, A ij represents the connection strength between node i and node j, k i and k j They represent the degrees of node i and node j respectively, m represents the total connection strength of the network, and c i and c j denote the communities to which nodes i and j belong respectively, δ(c i ,c j ) is the Kroneckerdelta function, when c i =c j The value is 1 when , otherwise it is 0.

7. The early warning method based on tunnel historical event feature analysis according to claim 6 is characterized in that: The event feature extraction and model building in S4 include: S41, extract key features: based on the analysis results of the event association network, extract key features from the historical event data. The key features include environmental conditions, geographical location, duration, impact range, traffic flow, traffic operation status, and lane closure when the event occurred; S42, establishing a historical event feature model: establishing a historical event feature model by combining the extracted key features with relevant information of the historical event data, and training the historical event feature model based on the historical event data to predict the probability and impact of the event; S43, model optimization: comprehensively consider the impact of various factors on tunnel safety, including geological disasters, traffic accidents, seismic activities, climate conditions and geographical location, and optimize the parameters of the historical event characteristic model.

8. The early warning method based on tunnel historical event feature analysis according to claim 7 is characterized in that: The historical event feature model adopts a Bayesian network model, and the Bayesian network model includes: Define the nodes of the Bayesian network: Each node represents a historical event type, including geological disasters, traffic accidents, earthquake activities, temperature changes, and water level changes. Each historical event node contains multiple attributes related to the historical event, including time, location, and severity; Determine the dependencies between nodes: Define the dependencies between nodes based on historical event data and domain knowledge, and use the dependencies to build a network structure, where edges represent conditional dependencies between historical events; Parameter learning and Bayesian network construction: In a Bayesian network, the behavior of each node depends on its parent node. The dependency is expressed through a conditional probability table. For node X i Its parent node set Pa(X i ), conditional probability P(X i |Pa(X i )) indicates that node X is in the state of the parent node when it is known i The state probability of node X i The state probability of is estimated by the statistical information in the data set, and the calculation formula is: Among them, N(x i ,pa(x i )) indicates that node X in the data set i and its parent node set pa(x i ) is the number of observations that are in the same state at the same time, N(pa(x i )) is the parent node set pa(x i ) the number of observations in that state; Network reasoning and prediction: The reasoning of Bayesian networks uses the belief propagation algorithm to calculate the edge probability of any node in the network. The calculation formula is: Among them, Bel(X i ) is node X i The belief is from node X j To Node X i The message, α is the normalization constant; From Node X j To Node X i The calculation formula for the message is: Among them, P(X j ∣X i ,Pa(X j )) indicates that when X is known i and X j Other parent nodes Pa(X j ), X j The conditional probability of .

9. The early warning method based on tunnel historical event feature analysis according to claim 8 is characterized in that: The real-time monitoring data fusion and analysis in S5 includes: S51, data fusion: real-time monitoring data is integrated with historical event feature models to form a comprehensive data view, and the real-time monitoring data is standardized and normalized; S52, anomaly detection and threshold setting: Calculate the anomaly detection score and set the threshold of the anomaly detection score based on experience or statistical analysis. When the anomaly detection score of the real-time monitoring data exceeds the set threshold, it is considered a risk event. The calculation formula of the anomaly detection score is: Among them, x is the data point monitored in real time, μ and σ are the mean and standard deviation calculated based on historical event data; S53, early warning rules: formulate early warning rules based on historical data and expert experience; S54, risk assessment and response: Combine real-time monitoring data and historical event feature models, use a linear regression model to conduct event risk assessment, and automatically initiate corresponding early warning mechanisms based on the results of the risk assessment, including activating traffic control signals and issuing public safety notices. The linear regression model is expressed as: y=β0+β1x1+β2x2+…+β n x n +∈; Among them, y is the response variable, x1, x2, …, x n are predictive variables of various factors affecting the occurrence of events, β0, β1, …, β n are model parameters and ∈ is the error term.

10. An early warning system based on characteristic analysis of tunnel historical events, used to implement an early warning method based on characteristic analysis of tunnel historical events as claimed in any one of claims 1 to 9, characterized in that: Includes the following modules: Historical event data collection and preprocessing module: responsible for collecting historical event data inside the tunnel, and cleaning and standardizing the historical event data; Historical event classification and summarization module: automatically classifies the collected historical event data and establishes an event type database; Event association network and data analysis module: Based on the classification results of the historical event classification and induction module, an event association network of tunnel historical events is established to analyze and reveal the implicit relationship and impact transmission path between events; Event feature extraction and model building module: extract key features from historical event data based on the analysis results of the event association network, combine the extracted key features to build a historical event feature model, and comprehensively consider the impact of various factors on tunnel safety; Real-time monitoring data fusion and analysis module: The real-time monitoring data is integrated with the established historical event feature model to achieve comprehensive monitoring and analysis of the internal status of the tunnel. Through anomaly detection algorithms and early warning rules, abnormal situations are discovered, and based on the degree of match between the real-time monitoring data and the historical event feature model, it is determined whether there is a risk inside the tunnel and the corresponding early warning mechanism is triggered.

Images