A city operation risk early warning method and system based on big data analysis

Abstract

The application discloses a kind of city operation risk early warning method and system based on big data analysis, it is related to city road monitoring technical field, including, collection city road data and vehicle data, extract vehicle health characteristics, define road network node and neighborhood range, calculate the risk heat map of vehicle health density around node;Build road network damage inversion neural network, generate road network node comprehensive vector, build road network node topology graph, use inversion layer to map road network node comprehensive vector as road damage type, build road chain reaction by dynamic bayesian failure chain propagation tree modeling, quantize and predict future city road risk;Using multi-strategy intervention simulation engine, generate intervention scheme based on predicted road risk, visualize early warning city road risk and record intervention scheme.The application accurately quantizes and predicts future city road risk, realizes real-time early warning and accurate intervention to road risk.

Description

Technical Field

[0001] This invention relates to the field of urban road monitoring technology, and in particular to a method and system for early warning of urban operation risks based on big data analysis. Background Technology

[0002] With the acceleration of urbanization, urban transportation systems are becoming increasingly complex, and the challenges of maintaining and managing transportation facilities are becoming increasingly severe. Especially in modern cities, factors such as traffic accidents, aging facilities, and extreme weather are having a growing impact on road safety and smooth traffic flow. This places higher demands on the stability of urban operations and traffic management. In recent years, with the rapid development of big data technology and artificial intelligence, big data-based urban road risk early warning systems have gradually become a research and application hotspot. By collecting, analyzing, and processing large amounts of data in urban transportation systems, early warning of potential road risks and precise deployment of risk prevention and control measures can be achieved. For example, the integration of urban road data, vehicle data, and traffic accident information, through data mining and machine learning technologies, can analyze road health conditions, traffic flow, and areas with frequent traffic accidents, thereby predicting and assessing potential risk points.

[0003] Existing technologies in urban road risk early warning systems mainly focus on the monitoring and analysis of single factors, often emphasizing the impact of road infrastructure damage, traffic flow, or weather conditions on traffic risks. They fail to fully consider the relationship between road networks and vehicle health characteristics and lack real-time, big data-based dynamic optimization capabilities. Summary of the Invention

[0004] In view of the aforementioned existing problems, the present invention is proposed.

[0005] Therefore, this invention provides a method and system for early warning of urban operation risks based on big data analysis, which solves the problem that existing technologies fail to fully consider the relationship between road networks and vehicle health characteristics and lack real-time, big data-based dynamic optimization capabilities.

[0006] To solve the above-mentioned technical problems, the present invention provides the following technical solution:

[0007] In a first aspect, the present invention provides a method for early warning of urban operational risks based on big data analysis, comprising,

[0008] Collect urban road and vehicle data, extract vehicle health characteristics, define road network nodes and neighborhood ranges, calculate vehicle health density around nodes, and generate risk heat maps.

[0009] A road network damage inversion neural network is constructed to generate a comprehensive vector of road network nodes. A road network node topology graph is constructed, and an inversion layer is used to map the comprehensive vector of road network nodes to road damage types. A dynamic Bayesian failure chain propagation tree is used to model road chain reactions, quantify and predict future urban road risks.

[0010] A multi-strategy intervention simulation engine is used to generate intervention plans based on predicted road risks, visualize and warn of urban road risks, and record intervention plans.

[0011] As a preferred embodiment of the urban operation risk early warning method based on big data analysis described in this invention, the collection of urban road data and vehicle data refers to obtaining vehicle data from the vehicle's ECU using the OBD interface, including vehicle sensor data and fault data, and simultaneously collecting the vehicle's driving trajectory through the vehicle's GPS module, thereby obtaining urban road attribute information through the vehicle's driving trajectory.

[0012] As a preferred embodiment of the urban operation risk early warning method based on big data analysis described in this invention, the steps of extracting vehicle health features, defining road network nodes and their neighborhood ranges, calculating the vehicle health density around the nodes, and generating a risk heatmap refer to defining urban road nodes based on urban road topology information, generating an urban road node map based on urban road connectivity, defining a neighborhood range for each urban road node, statistically analyzing vehicle data within the neighborhood range, extracting fault data from the vehicle data, calculating the proportion of each type of fault data in the vehicle data as the fault proportion, and normalizing and weighting the fault proportions of each type of fault data to generate the health density h of the urban road nodes. i ;

[0013] Calculate the average health density μ of all urban road nodes, and calculate the deviation D(i) for each urban road node;

[0014] A deviation matrix is ​​generated based on the deviations of all urban road nodes. The deviation values ​​in the deviation matrix are normalized. Urban road nodes with deviation values ​​less than a preset deviation threshold are marked as risk. The normalized deviation matrix is ​​converted into a risk heat map using a visualization tool, and the risk markers are simultaneously displayed on the risk heat map.

[0015] As a preferred embodiment of the urban operation risk early warning method based on big data analysis described in this invention, the method involves: constructing a road network damage inversion neural network to generate a comprehensive vector of road network nodes, constructing a road network node topology map, using an inversion layer to map the comprehensive vector of road network nodes to road damage type indicators, extracting risk-marked urban road nodes from the risk heatmap as road network nodes, extracting vehicle data from the neighborhood of the road network nodes, preprocessing it, and concatenating it to form a vehicle health feature vector V. iSimultaneously acquire road attribute information of road network nodes, preprocess and splice it to form road feature vector F i The vehicle health feature vector V of the road network node i and road feature vector F i The data is spliced ​​together to form a comprehensive vector X for road network nodes. i ;

[0016] DBSCAN clustering is used to synthesize the road network node vector X. i Cluster analysis is performed to obtain clustering results. Based on the clustering results, the road network nodes are divided into sets N(i) of road network nodes. For each set of road network nodes, the comprehensive vector X of the road network nodes is calculated. i The cosine similarity is used to determine whether there is a connection edge between road network nodes when the cosine similarity is greater than the similarity threshold. Based on the road network nodes and the connection edges, a road network node topology graph is formed for each set of road network nodes.

[0017] The graph network node topology graph is updated using graph convolution operations based on graph neural networks;

[0018] A road network damage inversion neural network is constructed using a fully connected neural network. The input is the updated node feature representation generated by the graph neural network, and the output is the probability distribution P of the damage type. i , is represented as;

[0019] Based on the damage probability distribution of road network nodes, the damage type of the road network nodes is determined by the damage threshold and output;

[0020] The probability distribution of damage types of road network nodes is used as the damage vector. The damage vectors of all road network nodes in the set of road network nodes are summarized to analyze the main damage types of urban road network nodes.

[0021] As a preferred embodiment of the urban operation risk early warning method based on big data analysis described in this invention, the method of modeling road chain reactions through dynamic Bayesian failure chain propagation tree, quantifying and predicting future urban road risks refers to taking the damage type of each road network node as a Bayesian network node, determining the causal relationship between nodes and representing it using a causal graph, and using the Bayesian network propagation mechanism to infer the node propagation probability through conditional probability.

[0022] Calculate the real-time risk value R of road network nodes using node propagation probability. t ;

[0023] An LSTM model is constructed as a risk prediction model, with the real-time risk value R of the road network nodes as input. t The output is the future urban road risk. The risk prediction model is trained and optimized using a loss function and the Adam optimizer, and the real-time risk value R is obtained. tThe trained risk prediction model is input to obtain the future urban road risks, and the damage type with the highest propagation probability is selected as the cause of the future urban road risks based on the node propagation probability of Bayesian network inference.

[0024] As a preferred embodiment of the urban operation risk early warning method based on big data analysis described in this invention, the method of generating intervention schemes based on predicted road risks using a multi-strategy intervention simulation engine refers to using a multi-strategy intervention simulation engine to predefine intervention scheme templates based on future urban road risks and the causes of future urban road risks, creating a digital twin model of urban roads based on vehicle information and urban road attribute information, inputting the intervention scheme into the urban road digital twin model for real-time simulation, calculating the real-time risk value of the city after simulation, and selecting the intervention scheme that minimizes the real-time risk value as the final intervention scheme.

[0025] As a preferred embodiment of the urban operation risk early warning method based on big data analysis described in this invention, the method of visually warning urban road risks and recording intervention plans refers to marking the predicted future urban road risks of each road network node on the risk heat map, simultaneously displaying the main damage types of urban road network nodes, and recording and storing the final intervention plan content in the database.

[0026] Secondly, this invention provides an urban operation risk early warning system based on big data analysis, comprising:

[0027] The data acquisition and analysis module is used to collect urban road data and vehicle data, extract vehicle health characteristics, define road network nodes and neighborhood ranges, calculate vehicle health density around nodes, and generate risk heat maps.

[0028] The risk prediction module is used to construct a road network damage inversion neural network, generate a comprehensive vector of road network nodes, construct a road network node topology map, use an inversion layer to map the comprehensive vector of road network nodes to road damage types, model road chain reactions through a dynamic Bayesian failure chain propagation tree, and quantify and predict future urban road risks.

[0029] The intervention recording module is used to generate intervention plans based on predicted road risks using a multi-strategy intervention simulation engine, visualize and warn of urban road risks, and record intervention plans.

[0030] Thirdly, the present invention provides a computer device including a memory and a processor, wherein the memory stores a computer program, wherein when the computer program is executed by the processor, it implements any step of the urban operation risk early warning method based on big data analysis as described in the first aspect of the present invention.

[0031] Fourthly, the present invention provides a computer-readable storage medium having a computer program stored thereon, wherein: when the computer program is executed by a processor, it implements any step of the urban operation risk early warning method based on big data analysis as described in the first aspect of the present invention.

[0032] The beneficial effects of this invention are as follows: This invention generates a risk heat map by calculating the vehicle health density around the node, effectively identifying high-risk areas in the urban road system and extracting risky road network nodes. Furthermore, it adopts a road network damage inversion neural network and models the chain reaction of road facilities through a dynamic Bayesian failure chain propagation tree, accurately quantifying and predicting future urban road risks, and realizing real-time early warning and precise intervention for road risks. Attached Figure Description

[0033] To more clearly illustrate the technical solutions of the embodiments of the present invention, the drawings used in the following description of the embodiments will be briefly introduced. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0034] Figure 1 This is a flowchart of the urban operation risk early warning method based on big data analysis in Example 1.

[0035] Figure 2 This is a structural diagram of the urban operation risk early warning system based on big data analysis in Example 1. Detailed Implementation

[0036] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the specific embodiments of the present invention will be described in detail below with reference to the accompanying drawings.

[0037] Many specific details are set forth in the following description in order to provide a full understanding of the invention. However, the invention may also be practiced in other ways different from those described herein, and those skilled in the art can make similar extensions without departing from the spirit of the invention. Therefore, the invention is not limited to the specific embodiments disclosed below.

[0038] Secondly, the term "one embodiment" or "embodiment" as used herein refers to a specific feature, structure, or characteristic that may be included in at least one implementation of the present invention. The phrase "in one embodiment" appearing in different places in this specification does not necessarily refer to the same embodiment, nor is it a single or selective embodiment that is mutually exclusive with other embodiments.

[0039] Example 1, referring to Figure 1 and Figure 2This is the first embodiment of the present invention, which provides a method for early warning of urban operational risks based on big data analysis, including the following steps:

[0040] S1. Collect urban road data and vehicle data, extract vehicle health characteristics, define road network nodes and neighborhood ranges, calculate vehicle health density around nodes, and generate risk heat maps.

[0041] Specifically, collecting urban road data and vehicle data refers to using the OBD interface to obtain vehicle data from the vehicle's ECU, including vehicle sensor data and fault data. Fault data includes engine failure, abnormal brake temperature, low tire pressure, etc. Simultaneously, the vehicle's driving trajectory is collected through the vehicle's GPS module, and urban road attribute information, including road slope, radius of curvature, etc., is obtained through the vehicle's driving trajectory.

[0042] Furthermore, vehicle health features are extracted, road network nodes and their neighborhood ranges are defined, and the vehicle health density around each node is calculated to generate a risk heatmap. This involves defining urban road nodes based on urban road topology information, generating an urban road node map based on urban road connectivity, defining a neighborhood range for each urban road node (e.g., a radius of 500 or 1000 meters), statistically analyzing vehicle data within this neighborhood, and extracting fault data from the vehicle data. The proportion of each fault type in the vehicle data is calculated as the fault ratio. For example, if 5 out of 20 vehicles in the neighborhood of a node experience engine failure, the engine failure rate of that node is 5 / 20 = 0.25. The fault ratios of each type of fault data are then normalized and weighted, and the sum is calculated to generate the health density h of the urban road node. i :

[0043] h i =1-m;

[0044] Where m is the normalized weighted sum of the fault proportions;

[0045] Calculate the average health density μ for all urban road nodes, and calculate the deviation D(i) for each urban road node:

[0046] D(i)=h i -μ;

[0047] A deviation matrix is ​​generated based on the deviations of all urban road nodes. Each element of the matrix represents the deviation value of a node. The deviation values ​​in the deviation matrix are normalized. Urban road nodes with deviation values ​​less than a preset deviation threshold are marked as risk. A visualization tool is used to convert the normalized deviation value matrix into a risk heat map. Urban road nodes with deviation values ​​greater than or equal to 0 are marked in green, and urban road nodes with deviation values ​​less than 0 are marked in red. The larger the absolute value of the deviation value, the darker the color. The risk is simultaneously marked on the risk heat map.

[0048] By collecting vehicle data and extracting vehicle health characteristics, the health density of each urban road node is calculated, which helps to obtain an intuitive reflection of the vehicle health status across the entire city. By calculating the health density of each node and comparing it with the city's average, nodes with poor health status can be effectively identified. By generating a deviation matrix for each node, the bias in risk assessment due to a single fault type is effectively avoided, enabling the system to comprehensively reflect the differences between different nodes in the road network. This risk assessment method based on deviation values ​​has extremely high sensitivity and accuracy, and can promptly identify high-risk nodes that are easily overlooked. By using visualization tools to convert the normalized deviation value matrix into a risk heat map, the health status of urban roads can be displayed intuitively and clearly.

[0049] S2. Construct a road network damage inversion neural network to generate a comprehensive vector of road network nodes, construct a road network node topology graph, use an inversion layer to map the comprehensive vector of road network nodes to road damage types, model road chain reactions through a dynamic Bayesian failure chain propagation tree, and quantify and predict future urban road risks.

[0050] Specifically, a road network damage inversion neural network is constructed to generate a comprehensive vector of road network nodes. A road network node topology graph is constructed, and an inversion layer is used to map the comprehensive vector of road network nodes to road damage type indicators. Risk-marked urban road nodes in the risk heatmap are extracted as road network nodes. Vehicle data in the neighborhood of the road network nodes are extracted, preprocessed, and concatenated to form a vehicle health feature vector V. i Simultaneously acquire road attribute information of road network nodes, preprocess and splice it to form road feature vector F i The vehicle health feature vector V of the road network node i and road feature vector F i The data is spliced ​​together to form a comprehensive vector X for road network nodes. i ;

[0051] DBSCAN clustering is used to synthesize the road network node vector X. i Cluster analysis is performed to obtain clustering results. Based on the clustering results, the road network nodes are divided into sets N(i) of road network nodes. For each set of road network nodes, the comprehensive vector X of the road network nodes is calculated. i The cosine similarity is used to determine whether there is a connection edge between road network nodes when the cosine similarity is greater than the similarity threshold. Based on the road network nodes and the connection edges, a road network node topology graph is formed for each set of road network nodes.

[0052] The road network node topology graph is updated using graph convolution operations based on graph neural networks.

[0053]

[0054] Where H k+1 For the node feature representation of the (k+1)th layer in the graph convolution operation, c iu W represents the cosine similarity between road network node i and road network node u. (k) Let σ be the weight matrix of the k-th layer, and σ be the activation function.

[0055] A road network damage inversion neural network is constructed using a fully connected neural network. The input is the updated node feature representation generated by the graph neural network, and the output is the probability distribution P of the damage type. i , is represented as:

[0056] P i =softmax(W out *H k +b out );

[0057] Where P i Let W be the probability distribution of the damage type of road network node i. out and b out The output weight matrix and bias term of the road network damage inversion neural network are given, where the damage types include road surface settlement, traffic light phase conflict, road cracks, etc.

[0058] Based on the damage probability distribution of road network nodes, the damage type of the road network nodes is determined by the damage threshold and output;

[0059] The probability distribution of damage types of road network nodes is used as the damage vector. The damage vectors of all road network nodes in the set of road network nodes are summarized to analyze the main damage types of urban road network nodes.

[0060] Concatenating vehicle health feature vectors with road feature vectors to form a comprehensive road network node vector helps to comprehensively consider multi-dimensional data within the same dimension. This allows for a precise description of the health status of each road network node from multiple perspectives. By integrating diverse data, the assessment of node status becomes more comprehensive, avoiding the bias of information. Furthermore, combining vehicle health data with road infrastructure data helps identify potentially high-risk areas, especially nodes where the risk is exacerbated by the combined effects of road damage and vehicle health issues. The use of cosine similarity provides an ideal method for assessing the similarity of road network nodes. Through clustering algorithms, the system can identify similar nodes and group them into the same group, effectively reducing the impact of data noise, especially when the dimensionality of vehicle health data and road infrastructure data is high. The system can avoid biases caused by inconsistencies in the dimensions of data across different dimensions. Through cluster analysis, the system can accurately group the road network, thereby forming a more reasonable topology map. This ensures that subsequent graph convolution operations can be performed on a more realistic structure, thus improving the accuracy of damage prediction. The combination of graph neural networks and graph convolution operations effectively utilizes the topology of urban road networks, realizing multi-level node feature transmission. It can capture complex interdependencies between nodes, thereby improving the ability to predict facility damage. Through the probability distribution of damage types, managers can obtain accurate risk predictions and provide decision support for subsequent maintenance and repair work. The neural network-based prediction method can automatically learn different types of facility damage patterns from the data, improving the accuracy and adaptability of prediction.

[0061] Furthermore, by modeling road cascading reactions using a dynamic Bayesian failure chain propagation tree, the future risks of urban roads are quantified and predicted. This involves treating the damage type of each road network node as a Bayesian network node, determining the causal relationships between nodes, and representing them using a causal graph. For example, roadbed settlement can lead to potholes in the road surface; these causal relationships can be determined using historical data, engineering expert knowledge, or survey data. The propagation probability of nodes is inferred using a conditional probability inference mechanism through the Bayesian network propagation mechanism.

[0062]

[0063] Where v j and v i These are the parent and child nodes in the Bayesian network, respectively, P(v i |v j Let P(v) be the probability of a child node appearing given a parent node. j |v i Let P(v) be the probability of a parent node appearing given a child node. i ) and P(v j) represent the occurrence probabilities of child nodes and parent nodes, respectively, which are extracted from the damage probability distribution of road network nodes;

[0064] Calculate the real-time risk value R of road network nodes using node propagation probability. t :

[0065]

[0066] Where w i Let be the risk weight for the i-th damage type, and n be the total number of damage types;

[0067] An LSTM model is constructed as a risk prediction model, with the real-time risk value R of the road network nodes as input. t The output is the future urban road risk. The risk prediction model is trained and optimized using a loss function and the Adam optimizer, and the real-time risk value R is obtained. t The trained risk prediction model is input to obtain the future urban road risks, and the damage type with the highest propagation probability is selected as the cause of the future urban road risks based on the node propagation probability of Bayesian network inference.

[0068] Bayesian networks can effectively infer causal relationships between nodes by combining historical data, expert knowledge, and survey data. Unlike traditional risk assessment methods based on a single data source, this invention constructs a more accurate and comprehensive damage propagation model through multi-dimensional information fusion. This model can capture the interaction between damages to different facilities. This method can provide more detailed and dynamic risk assessment for urban traffic management, helping to identify potential high-risk nodes and chain reactions in advance. By using the Bayesian network inference mechanism and updating the real-time risk value of road network nodes by calculating the node propagation probability, the risk level of different nodes can be dynamically adjusted, reflecting the propagation and changes of facility damage in real time. The LSTM model can extract time series information from real-time risk data and capture the long-term dependency relationship between facility damage and its chain reactions. This enables the system to not only assess current risks in real time but also effectively predict future road risks.

[0069] S3. Employ a multi-strategy intervention simulation engine to generate intervention plans based on predicted road risks, visualize and warn of urban road risks, and record intervention plans.

[0070] Specifically, the multi-strategy intervention simulation engine generates intervention plans based on predicted road risks. This means that the multi-strategy intervention simulation engine predefines intervention plan templates based on future urban road risks and their causes. These templates include, but are not limited to, temporary speed limits, signal timing adjustments, traffic control, and emergency repairs. A digital twin model of urban roads is created based on vehicle information and urban road attribute information. The intervention plans are then input into the digital twin model for real-time simulation. After the simulation, the real-time risk value of the city is calculated, and the intervention plan that minimizes the real-time risk value is selected as the final intervention plan.

[0071] By comprehensively considering multiple intervention strategies, complex urban traffic problems can be effectively addressed. For example, in some road sections, simply limiting speed may not be enough to solve traffic congestion; instead, adjusting traffic signal timing or implementing traffic control is necessary to optimize flow. The introduction of multiple strategies makes the system more flexible and adaptable, enabling it to quickly adjust and select the most appropriate intervention measures for different traffic conditions. The introduction of digital twin technology allows for real-time simulation of changes in urban road conditions in a virtual environment. This method has significant advantages: it can dynamically reflect the actual operating conditions of roads and evaluate the effectiveness of intervention measures through simulation. By establishing a digital twin model, the simulation engine can accurately capture the dynamic changes in urban roads, such as the impact of factors like road malfunctions, traffic density, and weather changes. By dynamically assessing risk values, the real-time effectiveness of each intervention measure can be ensured, avoiding the adoption of ineffective or inefficient response strategies. In urban traffic management, there are often differences in traffic conditions in different areas and at different times. Real-time risk value calculation can help accurately assess various risks, thereby helping managers formulate the most appropriate intervention strategies.

[0072] Furthermore, the visualization and early warning of urban road risks and the recording of intervention plans refer to marking the predicted future urban road risks of each road network node on the risk heat map, simultaneously displaying the main damage types of urban road network nodes, and recording and storing the final intervention plan content in the database.

[0073] This embodiment also provides an urban operation risk early warning system based on big data analysis, including:

[0074] The data acquisition and analysis module is used to collect urban road data and vehicle data, extract vehicle health characteristics, define road network nodes and neighborhood ranges, calculate vehicle health density around nodes, and generate risk heat maps.

[0075] The risk prediction module is used to construct a road network damage inversion neural network, generate a comprehensive vector of road network nodes, construct a road network node topology map, use an inversion layer to map the comprehensive vector of road network nodes to road damage types, model road chain reactions through a dynamic Bayesian failure chain propagation tree, and quantify and predict future urban road risks.

[0076] The intervention recording module is used to generate intervention plans based on predicted road risks using a multi-strategy intervention simulation engine, visualize and warn of urban road risks, and record intervention plans.

[0077] This embodiment also provides a computer device applicable to the urban operation risk early warning method based on big data analysis, including: a memory and a processor; the memory is used to store computer-executable instructions, and the processor is used to execute the computer-executable instructions to realize the urban operation risk early warning method based on big data analysis as proposed in the above embodiment.

[0078] The computer device can be a terminal, comprising a processor, memory, communication interface, display screen, and input devices connected via a system bus. The processor provides computing and control capabilities. The memory includes non-volatile storage media and internal memory. The non-volatile storage media stores the operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs stored in the non-volatile storage media. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, carrier networks, NFC (Near Field Communication), or other technologies. The display screen can be an LCD screen or an e-ink screen. The input devices can be a touch layer covering the display screen, buttons, a trackball, or a touchpad on the computer device's casing, or an external keyboard, touchpad, or mouse.

[0079] This embodiment also provides a storage medium storing a computer program, which, when executed by a processor, implements the urban operation risk early warning method based on big data analysis proposed in the above embodiments. The storage medium can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read Only Memory (EPROM), Programmable Red-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0080] In summary, this invention generates a risk heat map by calculating the vehicle health density around a node, effectively identifying high-risk areas in the urban road system and extracting risky road network nodes. Furthermore, it employs a road network damage inversion neural network and models the chain reaction of road facilities through a dynamic Bayesian failure chain propagation tree, accurately quantifying and predicting future urban road risks, thereby achieving real-time early warning and precise intervention for road risks.

[0081] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and are not intended to limit it. Although the present invention has been described in detail with reference to preferred embodiments, those skilled in the art should understand that modifications or equivalent substitutions can be made to the technical solutions of the present invention without departing from the spirit and scope of the technical solutions of the present invention, and all such modifications or substitutions should be covered within the scope of the claims of the present invention.

Claims

1. A method for early warning of urban operational risks based on big data analysis, characterized in that: include, Collect urban road and vehicle data, extract vehicle health characteristics, define road network nodes and neighborhood ranges, calculate vehicle health density around nodes, and generate risk heat maps. A road network damage inversion neural network is constructed to generate a comprehensive vector of road network nodes. A road network node topology graph is constructed, and an inversion layer is used to map the comprehensive vector of road network nodes to road damage types. A dynamic Bayesian failure chain propagation tree is used to model road chain reactions, quantify and predict future urban road risks. A multi-strategy intervention simulation engine is used to generate intervention plans based on predicted road risks, visualize and warn of urban road risks, and record intervention plans. The process involves constructing a road network damage inversion neural network to generate comprehensive vectors for road network nodes, building a road network node topology map, and using an inversion layer to map the comprehensive vectors of road network nodes to road damage type indicators. Risk-marked urban road nodes from the risk heatmap are extracted as road network nodes. Vehicle data from the neighborhood of each road network node is extracted, preprocessed, and concatenated to form a vehicle health feature vector. Simultaneously acquire road attribute information of road network nodes, preprocess and splice it to form road feature vectors. The vehicle health feature vector of the road network node and road feature vectors The data is spliced ​​together to form a comprehensive vector of road network nodes. ; DBSCAN clustering is used to synthesize the vector of road network nodes. Cluster analysis is performed to obtain clustering results. Based on the clustering results, the road network nodes are divided into sets of road network nodes. For each set of road network nodes, calculate the comprehensive vector of the road network nodes. The cosine similarity is used to determine whether there is a connection edge between road network nodes when the cosine similarity is greater than the similarity threshold. Based on the road network nodes and the connection edges, a road network node topology graph is formed for each set of road network nodes. The graph network node topology graph is updated using graph convolution operations based on graph neural networks; A road network damage inversion neural network is constructed using a fully connected neural network. The input is the updated node feature representation generated by the graph neural network, and the output is the probability distribution of damage types. , is represented as; Based on the damage probability distribution of road network nodes, the damage type of the road network nodes is determined by the damage threshold and output; The probability distribution of damage types of road network nodes is used as the damage vector. The damage vectors of all road network nodes in the set of road network nodes are summarized to analyze the main damage types of urban road network nodes. The method of modeling road chain reactions through dynamic Bayesian failure chain propagation tree, quantifying and predicting future urban road risks, refers to treating the damage type of each road network node as a Bayesian network node, determining the causal relationship between nodes and representing it using a causal graph, and using conditional probability inference of node propagation probability through the Bayesian network propagation mechanism. Calculate the real-time risk value of road network nodes using node propagation probability. ; An LSTM model is constructed as a risk prediction model, with the real-time risk values ​​of road network nodes as input. The output is the future urban road risk. The risk prediction model is trained and optimized using a loss function and the Adam optimizer, and the real-time risk value is used. The trained risk prediction model is input to obtain the future urban road risks, and the damage type with the highest propagation probability is selected as the cause of the future urban road risks based on the node propagation probability of Bayesian network inference.

2. The urban operation risk early warning method based on big data analysis as described in claim 1, characterized in that: The collection of urban road data and vehicle data refers to obtaining vehicle data from the vehicle's ECU using the OBD interface, including vehicle sensor data and fault data, and simultaneously collecting the vehicle's driving trajectory through the vehicle's GPS module to obtain urban road attribute information through the vehicle's driving trajectory.

3. The urban operation risk early warning method based on big data analysis as described in claim 2, characterized in that: The process of extracting vehicle health features, defining road network nodes and their neighborhoods, and calculating the vehicle health density around nodes to generate a risk heatmap involves defining urban road nodes based on urban road topology information, generating an urban road node graph based on urban road connectivity, defining a neighborhood for each urban road node, statistically analyzing vehicle data within that neighborhood, extracting fault data from the vehicle data, calculating the proportion of each type of fault data in the vehicle data as the fault ratio, and then normalizing and weighting the fault ratios of each type of fault data to generate the health density of the urban road nodes. ; Calculate the average health density of all urban road nodes. And calculate the deviation for each urban road node. ; A deviation matrix is ​​generated based on the deviations of all urban road nodes. The deviation values ​​in the deviation matrix are normalized. Urban road nodes with deviation values ​​less than a preset deviation threshold are marked as risk. The normalized deviation matrix is ​​converted into a risk heat map using a visualization tool, and the risk markers are simultaneously displayed on the risk heat map.

4. The urban operation risk early warning method based on big data analysis as described in claim 3, characterized in that: The aforementioned multi-strategy intervention simulation engine, which generates intervention plans based on predicted road risks, refers to using a multi-strategy intervention simulation engine to predefine intervention plan templates based on future urban road risks and their causes. It creates a digital twin model of urban roads based on vehicle information and urban road attribute information, inputs the intervention plan into the urban road digital twin model for real-time simulation, calculates the real-time risk value of the city after simulation, and selects the intervention plan that minimizes the real-time risk value as the final intervention plan.

5. The urban operation risk early warning method based on big data analysis as described in claim 4, characterized in that: The visualization and early warning of urban road risks and the recording of intervention plans refer to marking the predicted future urban road risks of each road network node on the risk heat map, simultaneously displaying the main damage types of urban road network nodes, and recording and storing the final intervention plan content in the database.

6. A city operation risk early warning system based on big data analysis, based on the city operation risk early warning method based on big data analysis as described in any one of claims 1 to 5, characterized in that: include, The data acquisition and analysis module is used to collect urban road data and vehicle data, extract vehicle health characteristics, define road network nodes and neighborhood ranges, calculate vehicle health density around nodes, and generate risk heat maps. The risk prediction module is used to construct a road network damage inversion neural network, generate a comprehensive vector of road network nodes, construct a road network node topology map, use an inversion layer to map the comprehensive vector of road network nodes to road damage types, model road chain reactions through a dynamic Bayesian failure chain propagation tree, and quantify and predict future urban road risks. The intervention recording module is used to generate intervention plans based on predicted road risks using a multi-strategy intervention simulation engine, visualize and warn of urban road risks, and record intervention plans.

7. A computer device comprising a memory and a processor, wherein the memory stores a computer program, characterized in that: When the processor executes the computer program, it implements the steps of the urban operation risk early warning method based on big data analysis as described in any one of claims 1 to 5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by the processor, it implements the steps of the urban operation risk early warning method based on big data analysis as described in any one of claims 1 to 5.

Images