Multi-protocol-driven pipe gallery risk identification and positioning method and system

Abstract

The invention discloses a pipe gallery risk identification and positioning method and system under multi-protocol driving, and relates to the technical field of pipe gallery risk management, and the method comprises the steps: obtaining a pipe gallery structure diagram, carrying out the deployment of a multi-source sensor, and constructing a sensor topology network; the pipe gallery is continuously monitored; characterization risk identification is carried out, characterization risk nodes and characterization risk feature vectors are determined, association link retrieval is carried out, and a target association link set is determined; extracting backtracking data, and determining a backtracking data sequence set; and carrying out serialized risk trend identification, determining an implicit risk node set, and taking the characterization risk node and the implicit risk node set as a risk identification positioning result. The technical problem of low risk monitoring efficiency caused by single sensor deployment and difficulty in accurate positioning of hidden risk nodes in pipe gallery risk monitoring in the prior art is solved, and the technical effects of realizing efficient identification and positioning of pipe gallery risks and effectively improving the pipe gallery risk monitoring and management and control capability are achieved.

Description

Technical Field

[0001] This application relates to the technical field of utility tunnel risk management, and particularly to a method and system for identifying and positioning utility tunnel risks driven by multiple protocols. Background Art

[0002] With the continuous expansion of the scale of urban underground utility tunnel construction, the complex environment and diverse facilities in the utility tunnel have put forward higher requirements for risk monitoring and control. Existing utility tunnel risk monitoring technologies generally have problems such as single sensor deployment, poor compatibility of data transmission protocols, and risk identification relying on manual experience, resulting in the inability to obtain real-time and comprehensive utility tunnel operation status data; at the same time, traditional methods can only identify surface risk characteristics and it is difficult to locate hidden risk nodes through historical data backtracking and risk trend analysis.

[0003] There are technical problems in existing utility tunnel risk monitoring, such as single sensor deployment, difficulty in accurately positioning hidden risk nodes, and low risk monitoring efficiency. Summary of the Invention

[0004] This application provides a method and system for identifying and positioning utility tunnel risks driven by multiple protocols, which are used to solve the technical problems in existing technologies, such as single sensor deployment in utility tunnel risk monitoring, difficulty in accurately positioning hidden risk nodes, and low risk monitoring efficiency.

[0005] In view of the above problems, this application provides a method and system for identifying and positioning utility tunnel risks driven by multiple protocols.

[0006] In the first aspect of this application, a method for identifying and positioning utility tunnel risks driven by multiple protocols is provided. The method includes: Obtain a utility tunnel structure diagram, deploy multi-source sensors in the utility tunnel based on the utility tunnel structure diagram to construct a sensor topology network; use the sensor topology network to continuously monitor the utility tunnel, and transmit the monitoring data to the transmission center of the sensor topology network through multiple communication protocols to generate a real-time sensor monitoring data topology network, where the transmission center includes an intermediate transfer and backtracking database; based on the real-time sensor monitoring data topology network, perform surface risk identification, determine surface risk nodes and surface risk feature vectors, retrieve associated links in the sensor topology network to determine a set of target associated links; for each associated topology node in the set of target associated links, extract backtracking data in the intermediate transfer and backtracking database to determine a set of associated topology node backtracking data sequences; perform serialized risk trend identification on the set of associated topology node backtracking data sequences to determine a set of hidden risk nodes, and use the surface risk nodes and the set of hidden risk nodes as the risk identification and positioning results.

[0007] In the second aspect of the present application, a risk identification and positioning system for utility tunnels under multi-protocol co-driving is provided. The system includes: A sensor topology network construction module, configured to obtain a utility tunnel structure diagram, deploy multi-source sensors in the utility tunnel based on the utility tunnel structure diagram, and construct a sensor topology network; a topology network generation module, configured to continuously monitor the utility tunnel by using the sensor topology network, and transmit monitoring data to a transmission center of the sensor topology network through multiple communication protocols to generate a real-time sensor monitoring data topology network, wherein the transmission center includes an intermediate transfer and backtracking database; an associated link set determination module, configured to perform characterization risk identification based on the real-time sensor monitoring data topology network, determine characterization risk nodes and a characterization risk feature vector, perform an associated link search in the sensor topology network, and determine a target associated link set; a backtracking data extraction module, configured to extract backtracking data in the intermediate transfer and backtracking database for each associated topology node in the target associated link set to determine an associated topology node backtracking data sequence set; an implicit risk node set determination module, configured to perform serialized risk trend identification on the associated topology node backtracking data sequence set to determine an implicit risk node set, and use the characterization risk nodes and the implicit risk node set as a risk identification and positioning result.

[0008] One or more technical solutions provided in the present application have at least the following technical effects or advantages: Obtain a utility tunnel structure diagram, deploy multi-source sensors, and construct a sensor topology network; continuously monitor the utility tunnel by using the sensor topology network, and transmit monitoring data to a transmission center of the sensor topology network through multiple communication protocols to generate a real-time sensor monitoring data topology network; perform characterization risk identification based on the real-time sensor monitoring data topology network, determine characterization risk nodes and a characterization risk feature vector, perform an associated link search, and determine a target associated link set; extract backtracking data in the intermediate transfer and backtracking database for each associated topology node in the target associated link set to determine an associated topology node backtracking data sequence set; perform serialized risk trend identification to determine an implicit risk node set, and use the characterization risk nodes and the implicit risk node set as a risk identification and positioning result. The technical effect of realizing efficient identification and positioning of utility tunnel risks and effectively improving the risk monitoring and control capabilities of utility tunnels is achieved. Description of the Drawings

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present application. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0010] Figure 1 Schematic diagram of the process of a risk identification and location method for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application; Figure 2 Another schematic diagram of the process of a risk identification and location method for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application; Figure 3 Another schematic diagram of the process of a risk identification and location method for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application; Figure 4 Another schematic diagram of the process of a risk identification and location method for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application; Figure 5 Another schematic diagram of the process of a risk identification and location method for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application; Figure 6 Another schematic diagram of the process of a risk identification and location method for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application; Figure 7 Another schematic diagram of the process of a risk identification and location method for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application; Figure 8 Schematic diagram of the structure of a risk identification and location system for utility tunnels under multi - protocol co - driving provided by an embodiment of the present application.

[0011] Explanation of reference numerals: Sensor topology network construction module 10, topology network generation module 20, associated link set determination module 30, backtracking data extraction module 40, implicit risk node set determination module 50. Detailed implementation manners

[0012] The present application provides a risk identification and location method and system for utility tunnels under multi - protocol co - driving, aiming to solve the technical problem in the prior art that in the risk monitoring of utility tunnels, the sensor deployment is single, and it is difficult to accurately locate implicit risk nodes, resulting in low risk monitoring efficiency.

[0013] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Apparently, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present application.

[0014] As Figure 1 shown, the present application provides a risk identification and location method for utility tunnels under multi - protocol co - driving, and the method includes: Step S100: Obtain the structure diagram of the utility tunnel, deploy multi-source sensors in the utility tunnel based on the structure diagram of the utility tunnel, and construct a sensor topology network.

[0015] Specifically, first obtain the structure diagram of the utility tunnel that includes the spatial layout, route direction, and node distribution of the utility tunnel, and extract the route set of the utility tunnel and the position of the route intersection points based on the structural features; then differentially deploy multiple types of sensors in the utility tunnel according to different monitoring requirements: acoustical sensors are mainly deployed in areas prone to leakage such as pipeline interfaces and valves (to capture abnormal sound waves generated by gas leakage), optical sensors are deployed in structurally weak sections (such as turning points and material transition zones) (to collect image information such as cracks and deformations in real time), and gas sensors are densely deployed along the flammable and explosive gas transmission pipelines (to monitor the concentration changes of methane, carbon monoxide, etc.); at the same time, all-type composite sensor nodes are added at the route intersection points to form an initial sensor set covering key risk points. Finally, taking each sensor as a topological node, construct a sensor topology network with spatial coordinates according to the physical connection relationship of the utility tunnel routes. The link between nodes reflects the spatial adjacency of the monitoring area, laying a foundation for subsequent spatial correlation analysis of multi-source data.

[0016] Step S200: Use the sensor topology network to continuously monitor the utility tunnel, and transmit the monitoring data to the transmission center of the sensor topology network through multiple communication protocols to generate a real-time sensor monitoring data topology network, where the transmission center includes an intermediate transfer and trace-back database.

[0017] Specifically, the sensor topology network continuously monitors the utility tunnel at high frequencies through multiple types of sensors (acoustical, optical, gas), and collects multi-modal data such as sound waveforms, structural images, and gas concentrations in real time. After being encoded by the adapted communication protocols (such as ZigBee for transmitting images, LoRa for transmitting gas concentrations, and Wi-Fi for transmitting acoustical signals), various types of data are aggregated by the edge nodes to the transmission center equipped with the intermediate transfer and trace-back database. As a unified data hub, this database stores the original sensor data and topological association information according to a preset time window (such as the most recent 7 days / 30 days) to form a structured data pool. After the transmission center performs timestamp synchronization, spatial coordinate calibration, and format normalization processing on the data, it generates a real-time sensor monitoring data topology network based on the connection relationship of the topological network nodes. This network dynamically maps the physical space of the utility tunnel and has both real-time data display and historical data quick retrieval functions. Compared with the traditional decentralized storage mode, through pre-aggregation and index optimization, the intermediate transfer and trace-back database enables the subsequent risk analysis to directly batch extract the backtracking data sequence of the target node based on the associated link of the topological network without retrieving data from each sensor one by one (such as the acoustical-optical-gas data chain of a certain pipeline section in the past 24 hours), reducing the data extraction time from the minute level to the second level and significantly improving the efficiency of multi-source data fusion analysis.

[0018] Step S300: performing risk characterization identification based on the real-time sensor monitoring data topology network, determining risk characterization nodes and risk characterization feature vectors, performing associated link retrieval in the sensor topology network, and determining a target associated link set.

[0019] Specifically, based on the real-time sensor monitoring data topology network, each topological node is automatically scanned through the preset multi-source risk identification rules (acoustic rules: sound frequency / amplitude exceeds the threshold; optical rules: cracks or deformations are detected by convolution analysis of image feature vectors; gas rules: gas type / concentration exceeds the standard). The nodes that trigger anomalies are marked as risk nodes, and the corresponding feature vectors are extracted (such as the acoustic feature vector contains abnormal frequency components, the optical feature vector contains the proportion of crack pixels, and the gas feature vector contains the concentration value that exceeds the standard). Subsequently, with the risk node as the center, the associated backtracking radius is determined based on the backtracking radius matching table (feature-distance mapping table generated by clustering historical risk data) matching the risk feature vector. For example, a gas leakage risk corresponds to a 50-meter backtracking radius, and a structural crack corresponds to a 30-meter backtracking radius. Based on this radius, all pipeline corridor routes and nodes that are connected to the risk node and whose distance is less than or equal to the backtracking radius are retrieved in the sensor topology network to form a target associated link set containing potential risk transmission paths. For example, when a node with excessive gas concentration triggers a search, all sensor nodes and connection routes within 50 meters upstream and downstream of the pipeline are included in the associated link, providing an analysis scope for subsequent tracing of the risk source.

[0020] Step S400: For each associated topology node in the target associated link set, backtracking data is extracted from the intermediate flow backtracking database to determine a backtracking data sequence set of the associated topology node.

[0021] Specifically, for each associated topology node within the target associated link set (such as acoustic, optical, and gas sensor nodes representing the upstream and downstream pipelines of the risk node), through the efficient indexing mechanism of the intermediate transfer backtracking database, historical monitoring data is batch-extracted according to the node ID and time range (such as pushing back 7 days from the current moment). The extracted data includes time series sequences of multi-source sensors (such as the sound frequency / amplitude curve of the acoustic sensor, the crack development image sequence of the optical sensor, and the concentration change trend of the gas sensor), device operation status parameters (such as valve switching frequency and pipeline pressure fluctuation), and topological association metadata (such as node spatial coordinates and link distance). The data extraction process is based on a predefined backtracking data structure template (including timestamp, sensor type, monitoring value, and device status field) to ensure the alignment of multi-source data in the spatio-temporal dimension and form a structured set of backtracking data sequences for associated topology nodes. For example, in the associated link triggered by a gas leakage characterization risk node, the high-frequency monitoring data chain of all sensors in the pipeline section where the node is located in the past 48 hours can be quickly retrieved, providing continuous and complete dataset support for subsequent analysis of the risk evolution trend and avoiding data retrieval time-consuming and association analysis errors caused by decentralized storage.

[0022] Step S500: Perform serialized risk trend identification on the set of backtracking data sequences for the associated topology nodes, determine the set of implicit risk nodes, and use the characterization risk node and the set of implicit risk nodes as the risk identification and location result.

[0023] Specifically, identify implicit risks by serially analyzing the set of backtracking data sequences of associated topology nodes: use time series analysis algorithms (such as calculating the fluctuation coefficient with a sliding window and dynamically time-warping to match historical abnormal patterns) to perform multi-dimensional evaluation on the historical data of each node (such as the sound frequency fluctuation curve, the gas concentration change trend, and the structural image feature sequence). For nodes with normal real-time data but significant fluctuations (such as the standard deviation exceeding 20% of the mean), historical intermittent abnormal records (such as short-term gas concentration exceeding the standard), or predicted trends reaching the risk threshold (such as the acceleration of crack propagation rate), they are determined as implicit risk nodes. Finally, the characterization risk nodes directly triggering the anomaly are merged with the implicit risk nodes mined through trend analysis to form a risk identification and location result covering immediate risks and potential hazards, achieving precise control from single-point anomaly alarm to full-link risk prediction.

[0024] As Figure 2 shown, in a possible implementation manner, step S100 further includes: Step S110: Extract the set of pipe gallery routes and the set of pipe gallery route features based on the pipe gallery structure diagram.

[0025] Step S120: Perform multi-source sensor layout on the utility tunnel route set respectively according to the utility tunnel route feature set to obtain an initial layout sensor set.

[0026] Step S130: Extract the route intersection points in the utility tunnel route set, and deploy multi-source sensors at each route intersection point to obtain an intersection layout sensor set.

[0027] Step S140: Take the union of the initial layout sensor set and the intersection layout sensor set to obtain a layout sensor set.

[0028] Step S150: Take each layout sensor in the layout sensor set as a topological node, and connect each topological node according to the route direction based on the utility tunnel route set to generate the sensor topological network.

[0029] Specifically, first obtain an electronic or paper utility tunnel structure diagram containing the utility tunnel plane layout, spatial orientation, and structural parameters. Through image recognition technology (such as convolutional neural network CNN) or manual annotation, extract the spatial geometric information (such as start and end coordinates, route length, bending angle, etc.) of all utility tunnel routes from the structure diagram to form a utility tunnel route set. At the same time, extract its attribute features for each route, including but not limited to: the type of conveyed medium (such as gas, tap water, heat), pipeline material (such as steel pipe, concrete pipe), service life, design pressure / flow parameters, historical maintenance records, etc., to form a utility tunnel route feature set corresponding one-to-one with the route set.

[0030] Implement a differential sensor deployment strategy based on the utility tunnel route feature set: for pipelines conveying flammable and explosive media, deploy gas and acoustic sensors at key nodes such as valves and interfaces and at preset intervals along the path to form a leakage monitoring network; for sections prone to structural deformation, deploy optical and stress sensors along the way to capture crack propagation and stress changes in real time; in high-temperature and high-pressure pipeline areas, configure a monitoring chain composed of temperature and pressure sensors. By matching and mapping route features (such as the type of conveyed medium, material characteristics, risk level) with sensor functions, customize a multi-type sensor combination for each route, and finally form an initial layout sensor set covering all utility tunnel routes to achieve directional perception and data collection for different risk scenarios.

[0031] Identify all route intersections (such as pipe branch points, cross nodes, and intersections of pipelines with different media) from the set of pipe gallery routes through spatial coordinate analysis. These intersections are prone to high-risk areas due to complex structures, stress concentration, or media interaction. For each intersection, deploy a composite monitoring unit containing multiple types of sensors such as acoustic, optical, and gas sensors: Acoustic sensors are used to capture abnormal sound waves generated by fluid diversion or sudden pressure changes, optical sensors monitor the structural deformation and crack initiation of nodes, and gas sensors continuously monitor the gas concentration changes in the intersection area (such as gas leakage and diffusion). By deploying multi-source sensors at intersections, a set of cross-deployed sensors is formed. For example, the composite sensors deployed at a cross node of a pipe gallery can simultaneously collect vibration, image, and gas data of pipelines in four directions, realizing three-dimensional monitoring of complex nodes, making up for the monitoring blind spots of single-type sensors, and improving the comprehensiveness and accuracy of risk perception.

[0032] Merge the data of the initial sensor set covering the main body of the pipe gallery route and the cross-deployed sensor set deployed for route intersections. By removing duplicate nodes (such as sensors where the route endpoints coincide with intersections) and retaining all types of sensors, a complete set of deployed sensors is generated. This set integrates the monitoring nodes of the route main body and the key intersection monitoring nodes, ensuring both continuous coverage of the linear path of the pipe gallery and enhanced three-dimensional monitoring of high-risk nodes.

[0033] Define each sensor in the set of deployed sensors as a topological node and assign a unique identifier (such as node ID, spatial coordinates). Based on the spatial orientation and connection relationship of the set of pipe gallery routes, connect adjacent nodes through virtual links according to the start-end logic: For straight routes, connect the nodes in sequence to form a chain-like topology; for branch routes, construct a star-shaped branch structure centered on the intersection node; for circular routes, generate a closed circular topology. Record attributes such as the physical distance and media flow direction between nodes during the connection process, and finally generate a sensor topological network that completely maps the physical structure of the pipe gallery.

[0034] As Figure 3 shown, in a possible implementation manner, step S300 further includes: Step S310: Obtain a representation of the risk identification rule.

[0035] Step S320: Use the representation of the risk identification rule to perform risk identification on the real-time sensor monitoring data topology network, and obtain a representation of risk nodes and a representation of risk feature vectors.

[0036] Specifically, by integrating the anomaly determination logics of multi-source sensors, risk identification rules are obtained, which specifically include: for acoustic sensors, setting the normal threshold ranges for sound frequency and amplitude, and triggering anomaly identification when the thresholds are exceeded; for optical sensors, using a pre-set convolutional model of crack image features to define the feature vector matching rules for structural anomalies; for gas sensors, setting the concentration safety limit (such as 10% of the lower explosion limit of methane) as the anomaly determination condition according to the gas type. After the above rules are defined in a standardized manner, a rule set containing the risk identification logics of acoustic, optical, and gas types is formed, providing a unified determination basis for the risk screening of real-time data.

[0037] Based on the obtained risk identification rules, each node data in the real-time sensor monitoring data topology network is matched type by type: for the acoustic sensor node, extract the monitored sound frequency and amplitude values, and compare them with the threshold ranges in the rules. If exceeded, mark the node as an acoustic-type risk representation node, and encapsulate the frequency and amplitude values into a feature vector; for the optical sensor node, perform convolutional analysis on the real-time image through a crack image feature recognizer to generate an image feature vector. If the crack feature parameters in the vector exceed the pre-set values in the rules, mark it as an optical-type risk representation node, and the feature vector carries the convolutional analysis results; for the gas sensor node, monitor the gas type and concentration in real time. If the concentration reaches the safety limit set in the rules (such as methane concentration ≥ 10% of the lower explosion limit), mark it as a gas-type risk representation node, and the feature vector contains the gas type and concentration values. By executing multiple rules in parallel, the nodes that trigger anomalies and their multi-dimensional feature vectors are captured in real time, forming an initial risk identification result. For example, a certain node is marked as a gas-type risk representation node because the monitored methane concentration is 0.8% LEL (lower than the lower explosion limit but exceeding the warning value), and the feature vector is recorded as [methane, 0.8% LEL], providing a clear risk anchor point for subsequent associated link analysis.

[0038] As Figure 4 shown, in a possible implementation manner, step S310 further includes: Step S311: Obtain the acoustic risk identification rule. Among them, the acoustic risk identification rule is that when the sensor type is a sound sensor, extract the sound frequency and sound amplitude monitored by the sound sensor for anomaly identification. If an anomaly is identified, use the topology node where the sound sensor is located as the risk representation node, and embed the sound frequency and sound amplitude into an initially empty vector to construct a risk representation feature vector.

[0039] Step S312: Obtain the optical risk recognition rule. Specifically, when the sensor type is an optical sensor, extract the image information monitored by the optical sensor, perform image feature convolution analysis on the image information using a crack image feature recognizer to obtain an image feature vector, conduct anomaly recognition based on the image feature vector. If an anomaly is recognized, take the topological node where the optical sensor is located as the risk characterization node and the image feature vector as the risk characterization feature vector.

[0040] Step S313: Obtain the gas risk recognition rule. Specifically, when the sensor is a gas sensor, extract the gas type and gas concentration monitored by the gas sensor for anomaly recognition. If an anomaly is recognized, take the topological node where the gas sensor is located as the risk characterization node, and embed the gas type and gas concentration into an initially empty vector to construct a risk characterization feature vector.

[0041] Step S314: Summarize the acoustic risk recognition rule, optical risk recognition rule, and gas risk recognition rule to generate a risk characterization recognition rule.

[0042] Specifically, for the risk recognition of the sound sensor type, define the acoustic risk recognition rule: After the sound sensor collects the sound data in the pipe gallery in real time, automatically extract the sound frequency and amplitude parameters, and perform anomaly determination through a preset normal operation threshold range (such as frequency 20 - 2000Hz, amplitude ≤ 80dB). If the monitored value exceeds the threshold (such as detecting a high-frequency abnormal sound with a frequency of 3000Hz or a sudden vibration with an amplitude of 95dB), the anomaly recognition logic is triggered. Mark the topological node where the sensor is located as the acoustic risk characterization node, and embed the actually monitored sound frequency and amplitude values into an initially empty feature vector (such as the vector representation is [frequency value, amplitude value]) to form a risk characterization feature vector carrying specific anomaly parameters, providing data support in the acoustic dimension for subsequent risk type classification and correlation analysis.

[0043] For the risk identification of optical sensor types, an optical risk identification rule is constructed: When the optical sensor collects the real-time image of the corridor structure, it automatically extracts the image information and inputs it into a pre-trained crack image feature recognizer (constructed based on the convolutional neural network algorithm). The recognizer traverses the image through multiple convolutional kernels, extracts parameters such as edge contours, pixel distributions, and texture features, and generates an image feature vector containing indicators such as crack length, width, and pixel ratio (for example, the vector format is [crack length, width, pixel ratio]). A preset abnormal determination threshold is set (such as crack width ≥ 0.2mm or pixel ratio ≥ 5%). If the parameters in the image feature vector exceed the threshold, it is determined that there is a structural abnormality, and the topological node where the corresponding optical sensor is located is marked as an optical type characterization risk node, and the image feature vector is directly used as the characterization risk feature vector, providing a basis for visual and quantitative analysis of structural deformation for subsequent risk positioning.

[0044] For the risk identification of gas sensor types, a gas risk identification rule is formulated: When the gas sensor monitors the gas data in the corridor in real time, it automatically analyzes the gas type (such as methane, carbon monoxide, oxygen, etc.) and the corresponding concentration value, and compares it with the threshold preset by the industry safety standard (such as the lower explosion limit of methane is 5%LEL, and the warning threshold is set to 1%LEL). If it is detected that the concentration of the target gas exceeds the warning threshold (such as the methane concentration reaches 1.2%LEL) or an abnormal gas type appears (such as a combustible gas that is not the medium transported by the pipeline), it is determined that there is a gas leakage or pollution risk, and the topological node where the sensor is located is marked as a gas type characterization risk node, and the actually monitored gas type and concentration value are embedded into an initially empty feature vector (such as the vector representation is [gas type, concentration value]), forming a characterization risk feature vector carrying specific risk parameters, providing key data support for subsequent risk tracing and diffusion analysis in the gas dimension.

[0045] Integrate and summarize the risk identification rules for the three types of acoustic, optical, and gas risks to form a unified characterization risk identification rule. Specifically, for the sound sensor, determine the risk based on the abnormal threshold of sound frequency and amplitude and construct a feature vector; for the optical sensor, identify the risk through the result of image feature convolution analysis and the crack parameter threshold and output a feature vector; for the gas sensor, mark the risk node and generate a feature vector according to whether the gas type and concentration exceed the safety limit. After the three types of rules are fused, a standardized risk identification logic covering multi-source sensors is formed, enabling parallel detection of risks such as acoustic anomalies, structural deformations, and gas leaks in the real-time monitoring data topology network, and uniformly generating a characterization risk identification result containing risk types, node positions, and multi-dimensional feature parameters, providing a structured rule support for subsequent risk correlation analysis and precise positioning.

[0046] Continue as Figure 3 shown, in a possible implementation manner, step S300 further includes: Step S330: Extract the associated corridor routes from the sensor topology network based on the positions of the risk - characterized nodes to obtain a set of associated corridor routes.

[0047] Step S340: Perform back - tracking radius matching according to the risk - characterized feature vector to obtain the associated back - tracking radius.

[0048] Step S350: Extract the associated links from the set of associated corridor routes based on the risk - characterized nodes and the associated back - tracking radius to determine the target set of associated links.

[0049] Specifically, according to whether the risk - characterized node is at a route intersection, perform differential extraction of associated corridor routes for the sensor topology network: If the risk - characterized node is located at a non - route intersection (i.e., the middle section or endpoint of a single corridor route), directly extract the only corridor route where the node is located as the associated route. For example, if an optical risk - characterized node is located in the middle of the gas pipeline L3, the set of associated corridor routes at this time only contains L3; If the risk - characterized node is located at a route intersection (such as the intersection of two or more corridor routes), then through the analysis of the connection relationship of the topology network, extract all corridor routes passing through this intersection. For example, if an acoustic risk - characterized node is located at the cross - intersection of three routes L1, L2, and L3, the set of associated corridor routes at this time will contain all routes of L1, L2, L3 and their extension directions. Through this differential extraction strategy, accurately identify the physical path range where the risk node is located.

[0050] Analyze the risk - characterized feature vector through a pre - constructed back - tracking radius matching mechanism to obtain the corresponding associated back - tracking radius: First, call the matching table containing the prototypes of historical risk feature vectors and preset back - tracking radii (this table is trained and generated based on the feature vectors and risk propagation distances in the historical corridor risk logs. For example, the historical maximum associated distance corresponding to a certain type of gas leakage feature is 30 meters), calculate the similarity (such as cosine similarity) between the current risk - characterized feature vector and the aggregated historical risk - characterized feature vector prototypes in the table, and select the prototype with the highest similarity as the matching result. Subsequently, extract the prototype back - tracking radius corresponding to this prototype from the matching table as the associated back - tracking radius for this analysis. For example, if the current feature vector is [methane, 1.5%LEL], and the similarity with the prototype vector marked as gas leakage in the matching table reaches 92%, and its corresponding prototype back - tracking radius is 40 meters, then the associated back - tracking radius is determined to be 40 meters. This process realizes the dynamic adaptation of the back - tracking range for different risk types through historical data - driven feature matching, providing a scientific spatial retrieval threshold for subsequent associated link extraction.

[0051] Centered on the risk node representation, combined with the associated backtracking radius and the set of associated corridor routes, perform the extraction of associated links with spatial range limitation: First, use the spatial coordinates of the risk node representation as the reference point. In the extracted set of associated corridor routes (such as a path network including main routes and intersection routes), extend upstream and downstream along the route direction, and delimit the physical distance retrieval range according to the associated backtracking radius (such as 30 meters, 50 meters). Then traverse all topological nodes and connection links within this range, and extract the links directly or indirectly connected to the risk node representation to form a target associated link set including attributes such as the connection relationship between nodes and the medium flow direction. For example, if the risk node representation is located 100 meters from the gas pipeline L2 and the associated backtracking radius is 40 meters, then in L2 and its intersecting routes L1 and L3, extract all sensor nodes and pipeline connections within the range from 60 meters to 140 meters of L2, including the links at the L1 / L2 intersection and the L2 / L3 intersection. This set not only covers the direct impact area around the risk node but also extends to the neighboring paths that may be affected by risk propagation through route association, providing a structured link network model for subsequent backtracking data extraction and risk trend analysis to ensure that the potential paths of risk propagation can be comprehensively captured.

[0052] As Figure 5 shown, in a possible implementation manner, step S340 further includes: Step S341: Pre-build a backtracking radius matching table.

[0053] Step S342: Perform a similarity match between the risk feature vector representation and the aggregated historical risk feature vector prototype in the backtracking radius matching table, and use the aggregated historical risk feature vector prototype corresponding to the maximum similarity value as the matching result.

[0054] Step S343: Extract the prototype backtracking radius of the matching result from the backtracking radius matching table as the associated backtracking radius.

[0055] Specifically, a backtracking radius matching table is pre-constructed through a historical data-driven method. The specific process is as follows: First, historical utility tunnel risk logs are collected, from which historical representative risk feature vectors (covering parameters such as acoustic frequency / amplitude, optical image features, gas type / concentration, etc.) and the historical association distances between the corresponding historical representative risk nodes and implicit risk nodes are extracted. Subsequently, a clustering algorithm is used to aggregate the historical representative risk feature vectors of the same type to form K aggregated feature vector sets, and the historical association distances corresponding to each set are aggregated into K distance sets (such as taking the maximum value or average value). Then, a prototype is constructed for each aggregated feature vector set, and the feature prototype is determined by iteratively calculating the spatial density in the M-dimensional space (for example, by randomly selecting spatial points, constructing subspaces, and verifying the iteration direction). Finally, each feature prototype and its corresponding maximum historical association distance (i.e., the prototype backtracking radius) are stored in the table to form a mapping relationship between the aggregated historical representative risk feature vector prototype - prototype backtracking radius. For example, the high-frequency vibration feature prototype - 50 meters, the methane leakage feature prototype - 40 meters, providing historical data support for subsequent matching the backtracking radius according to the real-time risk feature vector.

[0056] Calculate the similarity between the currently obtained representative risk feature vector (such as [sound frequency 2500Hz, amplitude 90dB, methane concentration 1.2%LEL]) and each aggregated historical representative risk feature vector prototype in the backtracking radius matching table. First, standardize the feature vector (such as z-score standardization) to eliminate the influence of dimension. Then, use the cosine similarity algorithm to calculate the directional similarity between vectors, and at the same time combine the Euclidean distance to evaluate the spatial distance, and form a comprehensive similarity index through weighted fusion (for example, the cosine similarity weight is 0.6, and the Euclidean distance weight is 0.4). For high-dimensional feature vectors (such as vectors containing image features), first reduce the dimension to within 30 dimensions through principal component analysis (PCA) and then calculate. Traverse all prototypes in the matching table, record the maximum similarity value and the corresponding prototype ID. If the maximum similarity exceeds the preset threshold (such as 0.8), then use this prototype as the matching result; if it does not exceed the threshold, then start the fuzzy matching mechanism, select the closest prototype and trigger the manual review process. For example, the comprehensive similarity between the current vector and the pipeline vibration feature prototype is 0.92, which exceeds the threshold, then determine this prototype as the matching result, and directly map its corresponding prototype backtracking radius (such as 50 meters) as the associated backtracking radius to ensure that the intelligent prediction of the risk propagation range is consistent with the historical law.

[0057] According to the obtained matching results, retrieve the corresponding prototype backtracking radius from the pre-constructed backtracking radius matching table as the associated backtracking radius for the current risk analysis. Each aggregated historical representation risk feature vector prototype in the matching table is associated with a prototype backtracking radius statistically based on historical data (for example, the high-frequency abnormal vibration prototype corresponds to 50 meters, and the methane concentration exceeding the standard prototype corresponds to 40 meters). When the prototype with the highest similarity to the current representation risk feature vector is determined (such as matching the gas leakage prototype), directly extract the prototype backtracking radius corresponding to this prototype (such as 40 meters) as the spatial range threshold for subsequent associated link extraction. Through the mapping relationship of historical risk data, this process transforms the abstract feature vector into a specific physical distance parameter, providing a quantitative basis for accurately delimiting the risk impact range, ensuring that the associated link analysis not only covers potential risk propagation paths but also avoids meaningless range expansion, and improving the efficiency and accuracy of risk positioning.

[0058] As Figure 6 shown, in a possible implementation manner, step S341 further includes: Step S3411: Obtain a set of historical risk logs of the pipe gallery.

[0059] Step S3412: Extract the historical association distances between the historical representation risk nodes and the historical implicit risk nodes in the set of historical risk logs of the pipe gallery, as well as the historical representation risk feature vectors, to obtain a set of historical representation risk feature vectors and a set of historical association distances.

[0060] Step S3413: Aggregate the set of historical representation risk feature vectors of the same type to obtain K sets of aggregated historical representation risk feature vectors, where K is a positive integer.

[0061] Step S3414: Map and aggregate the set of historical association distances according to the K sets of aggregated historical representation risk feature vectors to obtain K sets of aggregated historical association distances.

[0062] Step S3415: Construct prototypes for the K sets of aggregated historical representation risk feature vectors to generate K aggregated historical representation risk feature vector prototypes.

[0063] Step S3416: Take the maximum values in the K sets of aggregated historical association distances as the K prototype backtracking radii.

[0064] Step S3417: Construct a backtracking radius matching table according to the K aggregated historical representation risk feature vector prototypes and the K prototype backtracking radii.

[0065] Specifically, by connecting to the tunnel operation and maintenance database or historical monitoring system, a collection of historical tunnel risk logs can be obtained in batches. These logs contain data on various risk events recorded during tunnel operation, including but not limited to the occurrence time, sensor node location, monitoring parameters, and subsequent risk development of risk types such as acoustic anomalies, structural cracks, and gas leaks.

[0066] A structured analysis of the historical pipeline corridor risk log collection is performed: For each risk log, the spatial coordinates of the historical risk-signaling node (i.e., the sensor node where the anomaly was first detected, such as the sound sensor node corresponding to an acoustic anomaly) and the historical implicit risk node (i.e., the node subsequently confirmed to be the risk source or associated influence, such as the optical sensor node corresponding to the pipeline damage point) are first extracted. By calculating the Manhattan distance or Euclidean distance between the two points, the historical correlation distance (e.g., 35 meters) is calculated, and all distance values are aggregated into a historical correlation distance set. Simultaneously, monitoring data corresponding to the historical risk-signaling nodes is extracted from the logs, and a historical risk feature vector is constructed based on the sensor type: for sound sensors, the sound frequency and amplitude values (e.g., [2500Hz, 90dB]) are extracted; for optical sensors, the vector parameters after convolution analysis of the crack image features are extracted (e.g., [crack length 5cm, width 0.3mm]); and for gas sensors, the gas type and concentration values (e.g., [methane, 1.8%LEL]) are extracted. Ultimately, a set of historical risk feature vectors containing multi-dimensional parameters is formed. The above extraction process provides a key quantitative data basis for subsequent risk feature clustering and backtracking radius calculation.

[0067] Clustering algorithms (such as K-means and DBSCAN) are used to perform unsupervised learning on a collection of historical risk feature vectors. Vectors with similar characteristic parameters are grouped into the same category, thereby aggregating similar risk features. The specific process is as follows: First, the multidimensional feature vectors are normalized (e.g., by standardizing the parameter ranges of each dimension) to eliminate dimensional differences. Then, by iteratively calculating the Euclidean distance or cosine similarity between vectors, vectors with similar distances in the feature space are clustered into K clusters (the value of K can be determined using the elbow rule or silhouette coefficient). Each cluster corresponds to a clustered set of historical risk feature vectors. For example, all acoustic feature vectors with frequencies above 2000Hz and amplitudes greater than 85dB are clustered into a high-frequency strong vibration cluster, and all gas feature vectors with methane concentrations exceeding 1% LEL are clustered into a gas leak cluster. This clustering approach abstracts massive amounts of historical risk feature data into K representative categories, providing structured data support for subsequent prototype construction and backtracking radius calculation.

[0068] Use the generated set of K aggregated historical representation risk feature vectors as a mapping index to classify and aggregate the set of historical association distances: Traverse each historical association distance and classify the distance value into the corresponding category according to the aggregation category to which the corresponding representation risk feature vector belongs. For example, if the feature vector corresponding to a historical association distance of 35 meters belongs to the high-frequency strong vibration cluster, then classify 35 meters into the distance set corresponding to this cluster. Through this mapping relationship, the original set of historical association distances is divided into K independent sets of aggregated historical association distances, and each set corresponds to a specific type of risk feature (such as the distance set {25 meters, 35 meters, 40 meters} corresponding to the high-frequency strong vibration cluster, and the gas leakage cluster corresponding to {30 meters, 45 meters, 50 meters}). This aggregation method realizes the correlation modeling between risk features and propagation distances, provides a classification and statistical basis for determining the backtracking radius of different risk types subsequently, and ensures that the defined backtracking range matches the propagation characteristics of risk types.

[0069] Construct a prototype for each set of aggregated historical representation risk feature vectors to generate a typical vector that can represent the characteristics of this category: For each set, first randomly select an initial point in its feature space, then construct a subspace with a preset radius, and calculate the mean vector of all vectors within the subspace as the new center point; Iterate this process, and determine the final center point by verifying the convergence of the iteration direction (such as the distance between the center points of two consecutive iterations is less than the threshold). This point is the feature prototype of this set. For example, for the set of feature vectors of the gas leakage cluster, the prototype vector obtained through iterative calculation is [methane, 1.5% LEL, sound frequency 2500 Hz, amplitude 90 dB]. For high-dimensional feature vectors (such as image features), use principal component analysis (PCA) to reduce the dimension to within 30 dimensions before performing prototype calculation to ensure computational efficiency. Finally, each aggregated set generates a unique feature prototype, forming K aggregated historical representation risk feature vector prototypes, providing a standardized reference template for subsequent risk matching.

[0070] Perform an extreme value extraction operation on each set of aggregated historical association distances: Traverse the K sets, and for each set (such as the distance set {30 meters, 45 meters, 50 meters} corresponding to the gas leakage cluster), directly select the maximum value (such as 50 meters) as the prototype backtracking radius corresponding to this set. This operation is based on the principle of conservative estimation of the risk propagation range, ensuring that the selected radius can cover the maximum influence range shown by this type of risk in historical cases. For example, if the set of historical association distances of the structural vibration cluster is {25 meters, 35 meters, 40 meters}, then its prototype backtracking radius is set to 40 meters. In this way, the K sets of aggregated historical association distances are transformed into K prototype backtracking radii, providing a quantitative basis for subsequent matching of backtracking radii according to real-time risk features, and ensuring that the spatial retrieval range during risk analysis is sufficient to cover potential associated risk nodes.

[0071] Associate the K aggregated historical representation risk feature vector prototypes with the corresponding K prototype backtracking radii to construct a structured backtracking radius matching table: establish key-value pairs with the feature prototypes as keys (Key) and the prototype backtracking radii as values (Value) to form a mapping relationship table of feature prototype - backtracking radius. For example, associate the gas leakage prototype vector [methane, 1.5%LEL, 2500Hz, 90dB] with 50 meters. This table uses a multi-dimensional index structure to optimize the query efficiency, sets composite indexes for different-dimensional feature parameters such as acoustics, optics, and gas, ensuring that real-time risk feature vectors can be quickly matched to the corresponding prototypes. The finally generated matching table is stored in an in-memory database (such as Redis) as the core retrieval basis for similarity matching, supporting risk backtracking radius queries with millisecond-level responses, and providing an accurate spatial range definition for the extraction of association links.

[0072] As Figure 7 shown, in a possible implementation, step S3415 further includes: Step S34151: Extract the first aggregated historical representation risk feature vector set from the K aggregated historical representation risk feature vector sets.

[0073] Step S34152: Map the first aggregated historical representation risk feature vector set into an M-dimensional space respectively to obtain a set of spatial points, where M is a positive integer, and the coordinates of each spatial point correspond to a first aggregated historical representation risk feature vector.

[0074] Step S34153: Randomly extract a first spatial point from the set of spatial points, and construct a first subspace according to a preset spatial radius.

[0075] Step S34154: Randomly extract a spatial point on the edge of the first subspace as a direction verification spatial point, and construct a direction verification subspace based on the preset spatial radius.

[0076] Step S34155: Determine whether the spatial density of the first subspace is less than or equal to the spatial density of the direction verification subspace. If so, use the direction from the first spatial point to the direction verification spatial point as the iteration direction, and continue to iterate the direction verification spatial point in accordance with the iteration direction until the preset number of iterations is satisfied, and use the first aggregated historical representation risk feature vector corresponding to the spatial point obtained in the last iteration as the first aggregated historical representation risk feature vector prototype.

[0077] Step S34156: Traverse the K aggregated historical representation risk feature vector sets for prototype construction to obtain the K aggregated historical representation risk feature vector prototypes.

[0078] Specifically, among the K sets of aggregated historical representation risk feature vectors generated by the clustering algorithm, the first set is extracted in a preset order (such as the natural order of the clustering results, the risk type priority order, etc.) as the target set for the current prototype construction, denoted as the first set of aggregated historical representation risk feature vectors.

[0079] Convert each feature vector in the first set of aggregated historical representation risk feature vectors into a coordinate point in the M-dimensional space to form a set of spatial points. The dimensions of the feature vector (such as acoustic frequency, amplitude, gas type concentration, etc.) directly correspond to the dimensions M of the space (for example, if the feature vector contains 3 parameters: sound frequency, amplitude, and gas concentration, then M = 3). The value of each parameter after being standardized (such as normalized to the interval [0, 1]) is used as the coordinate value of the spatial point in the corresponding dimension. For example, a certain feature vector is [sound frequency 2000Hz, amplitude 80dB, methane concentration 1.2%LEL], and after standardization, it is converted into a point (0.6, 0.7, 0.8) in the three-dimensional space. Through this mapping, the abstract risk feature parameters are transformed into the position distribution in the geometric space, which is convenient for subsequent determination of the feature prototype through spatial density analysis and provides an intuitive geometric model basis for the iterative calculation of prototype construction.

[0080] Randomly select a point from the set of spatial points (i.e., the point set after mapping the first set of aggregated historical representation risk feature vectors to the M-dimensional space) as the first spatial point. The preset spatial radius is a parameter preset for defining the subspace range (for example, it can be set to 10% of the maximum Euclidean distance after feature vector standardization). With this spatial point as the center and the preset spatial radius as the range, construct a spherical or hypercube-shaped first subspace in the M-dimensional space, which contains all spatial points whose distance from the first spatial point is less than or equal to the preset radius.

[0081] Determine the edge region of the first subspace (i.e., the set of points whose distance from the first spatial point is equal to the preset spatial radius), and randomly select a spatial point from this edge region as the direction verification spatial point. Then, with this direction verification spatial point as the center, use the same preset spatial radius to construct a direction verification subspace, which is also of spherical or hypercube structure and is used to evaluate the feature distribution density in the current direction. For example, if the center of the first subspace is (0.6, 0.7, 0.8), the preset radius is 0.2, and the coordinate of the edge point A is (0.8, 0.7, 0.8), then the direction verification subspace takes point A as the center, with a radius of 0.2, and contains all spatial points whose distance from point A is ≤0.2. This step provides a comparison benchmark for subsequent comparison of the spatial density in different directions and determination of the iterative direction of the feature prototype by selecting verification points at the edge of the subspace and constructing a new subspace, ensuring that the prototype can converge to the region with the densest feature distribution.

[0082] By comparing the spatial density of the first subspace with that of the direction verification subspace (i.e., the number of spatial points per unit volume), the iterative direction of the feature prototype is dynamically adjusted: First, calculate the density of the two subspaces: Spatial density = number of points in the subspace / subspace volume. If the density of the first subspace is less than or equal to the density of the direction verification subspace, it indicates that there is a denser feature distribution along the direction from the first spatial point to the direction verification spatial point. Set this direction as the iterative direction, and use the direction verification spatial point as the new center. Repeat the process of constructing the subspace, selecting the edge points, and calculating the density until the preset number of iterations (e.g., 10 times) is reached. Each iteration moves towards the region with higher density, ensuring that the prototype gradually approaches the core region of the feature distribution. For example, in the first iteration, the density of the first subspace is 5 points per unit volume, and the density of the direction verification subspace is 8 points per unit volume. Then the iterative direction is from the first spatial point to the direction verification spatial point, and the new center is set as this verification point. Reconstruct the subspace and repeat the judgment. When the iteration ends, take the feature vector corresponding to the spatial point obtained in the last iteration as the prototype of the first aggregated historical representation risk feature vector. If the density of the first subspace is greater than that of the direction verification subspace, abandon the current direction and randomly select a direction verification spatial point from the edge of the first subspace (retry at most the preset number of times, e.g., 5 times). If none of them meet the conditions, directly use the vector corresponding to the initial first spatial point as the prototype to ensure the robustness of the process. Through this density-guided iterative mechanism, the prototype can accurately reflect the core distribution of the aggregated feature set, improving the accuracy of subsequent risk matching.

[0083] After completing the prototype construction of the first aggregated historical representation risk feature vector set (i.e., the first set), select the remaining K - 1 aggregated sets in the preset order (such as the index order of the clustering results, the priority of risk types, etc.), and repeat the prototype construction process of steps S34151 to S34155: For each set, first map its feature vectors to the M-dimensional space to form a point set, and then determine the feature prototype of this set through random point selection, subspace construction, density comparison, and iterative calculation. For example, if K = 3, after completing the prototype construction of the gas leakage class set, perform the same operations on the structural vibration class and gas concentration anomaly class sets in sequence. After traversing, generate a set containing K prototypes, and each prototype corresponds to the core feature vector of a clustering cluster, such as the high-frequency strong vibration prototype, the methane leakage prototype, etc. These prototypes form the core retrieval keys of the backtracking radius matching table, ensuring that subsequent real-time risk feature vectors can quickly match the corresponding prototypes and backtracking radii based on historical clustering rules, providing an efficient feature comparison benchmark for the risk positioning of the pipe gallery under multi-protocol driving.

[0084] In a possible implementation manner, step S3415 further includes: When the spatial density of the first subspace is greater than that of the direction verification subspace, re-select the direction verification space points and record the number of re-selections. When the number of re-selections is greater than or equal to the preset number of selections, use the first aggregated historical representation risk feature vector corresponding to the first space point as the prototype of the first aggregated historical representation risk feature vector.

[0085] Specifically, if the spatial density of the first subspace is greater than that of the direction verification subspace, it indicates that the direction feature distribution where the currently selected direction verification space point is located is sparser. In this case, the direction will be abandoned and a new direction verification space point will be randomly selected from the edge of the first subspace, while recording the number of re-selections. For example, if the preset number of selections is 5 times, when the first comparison does not meet the condition, new points will be re-selected and subspaces will be reconstructed again for density comparison. If after 5 consecutive re-selections, the densities of all direction verification subspaces are still less than that of the first subspace, it is considered that the first space point is already in the region of the maximum local density, and the corresponding feature vector will be directly used as the prototype of the first aggregated historical representation risk feature vector. This mechanism ensures that the algorithm can still converge in the case of complex or scattered feature distributions by limiting the number of retries, avoiding infinite loops, and at the same time ensuring that the prototype construction result can reflect the core features of the data as much as possible, improving the reliability of subsequent risk matching.

[0086] Based on the same inventive concept as a method for identifying and locating risks in a pipe gallery under multiple protocol drives in the foregoing embodiment, as Figure 8 shown, the present application provides a system for identifying and locating risks in a pipe gallery under multiple protocol drives. The system in the embodiments of the present application and the method embodiments are based on the same inventive concept. Among them, the system includes: A sensor topology network construction module 10, configured to obtain a pipe gallery structure diagram, deploy multi-source sensors in the pipe gallery based on the pipe gallery structure diagram, and construct a sensor topology network.

[0087] A topology network generation module 20, configured to continuously monitor the pipe gallery using the sensor topology network, and transmit the monitoring data to the transmission center of the sensor topology network through multiple communication protocols to generate a real-time sensor monitoring data topology network, where the transmission center includes an intermediate transfer and backtracking database.

[0088] An associated link set determination module 30, configured to perform characterization risk identification based on the real-time sensor monitoring data topology network, determine characterization risk nodes and characterization risk feature vectors, and perform associated link retrieval in the sensor topology network to determine a target associated link set.

[0089] A backtracking data extraction module 40, configured to extract backtracking data in the intermediate transfer and backtracking database for each associated topology node in the target associated link set to determine an associated topology node backtracking data sequence set.

[0090] The implicit risk node set determination module 50 is configured to perform serialized risk trend identification on the associated topological node backtracking data sequence set, determine the implicit risk node set, and use the represented risk nodes and the implicit risk node set as the risk identification and positioning result.

[0091] Furthermore, the system is also used for the following functions: Extract the pipe gallery route set and the pipe gallery route feature set based on the pipe gallery structure diagram; perform multi-source sensor layout on the pipe gallery route set respectively according to the pipe gallery route feature set to obtain the initial layout sensor set; extract the route intersection points in the pipe gallery route set, and layout multi-source sensors at each route intersection point to obtain the intersection layout sensor set; perform the union operation on the initial layout sensor set and the intersection layout sensor set to obtain the layout sensor set; use each layout sensor in the layout sensor set as a topological node, and connect each topological node according to the route trend of the pipe gallery route set to generate the sensor topology network.

[0092] Furthermore, the system is also used for the following functions: Obtain the represented risk identification rule; use the represented risk identification rule to perform risk identification on the real-time sensor monitoring data topology network to obtain the represented risk nodes and the represented risk feature vectors.

[0093] Furthermore, the system is also used for the following functions: Obtain the acoustic risk identification rules. Among them, the acoustic risk identification rules are as follows: when the sensor type is a sound sensor, extract the sound frequency and sound amplitude monitored by the sound sensor for anomaly identification. If an anomaly is identified, regard the topological node where the sound sensor is located as the risk characterization node, embed the sound frequency and sound amplitude into an initially empty vector, and construct a risk characterization feature vector; obtain the optical risk identification rules. Among them, the optical risk identification rules are as follows: when the sensor type is an optical sensor, extract the image information monitored by the optical sensor, perform image feature convolution analysis on the image information using a crack image feature recognizer to obtain an image feature vector, perform anomaly identification based on the image feature vector. If an anomaly is identified, regard the topological node where the optical sensor is located as the risk characterization node, and use the image feature vector as the risk characterization feature vector; obtain the gas risk identification rules. Among them, the gas risk identification rules are as follows: when the sensor is a gas sensor, extract the gas type and gas concentration monitored by the gas sensor for anomaly identification. If an anomaly is identified, regard the topological node where the gas sensor is located as the risk characterization node, embed the gas type and gas concentration into an initially empty vector, and construct a risk characterization feature vector; summarize the acoustic risk identification rules, optical risk identification rules, and gas risk identification rules to generate risk characterization identification rules.

[0094] Further, the system is also used for the following functions: Extract the associated corridor routes for the sensor topology network based on the positions of the risk characterization nodes to obtain a set of associated corridor routes; perform backtracking radius matching according to the risk characterization feature vectors to obtain the associated backtracking radius; perform associated link extraction on the set of associated corridor routes based on the risk characterization nodes and the associated backtracking radius to determine the target set of associated links.

[0095] Further, the system is also used for the following functions: Pre-construct a backtracking radius matching table; perform similarity matching between the risk characterization feature vectors and the aggregated historical risk characterization feature vector prototypes in the backtracking radius matching table, and regard the aggregated historical risk characterization feature vector prototype corresponding to the maximum similarity value as the matching result; extract the prototype backtracking radius of the matching result from the backtracking radius matching table as the associated backtracking radius.

[0096] Further, the system is also used for the following functions: Obtain the set of historical utility tunnel risk logs; extract the historical association distances between the historical representative risk nodes and the historical implicit risk nodes in the set of historical utility tunnel risk logs, as well as the historical representative risk feature vectors, to obtain a set of historical representative risk feature vectors and a set of historical association distances; perform homogeneous aggregation on the set of historical representative risk feature vectors to obtain K sets of aggregated historical representative risk feature vectors, where K is a positive integer; perform mapping aggregation on the set of historical association distances according to the K sets of aggregated historical representative risk feature vectors to obtain K sets of aggregated historical association distances; construct prototypes for the K sets of aggregated historical representative risk feature vectors to generate K prototypes of aggregated historical representative risk feature vectors; use the maximum value in the K sets of aggregated historical association distances as the K prototype backtracking radii; construct a backtracking radius matching table according to the K prototypes of aggregated historical representative risk feature vectors and the K prototype backtracking radii.

[0097] Further, the system is also used for the following functions: Extract the first set of aggregated historical representative risk feature vectors from the K sets of aggregated historical representative risk feature vectors; map the first set of aggregated historical representative risk feature vectors into an M-dimensional space respectively to obtain a set of space points, where M is a positive integer, and the coordinates of each space point correspond to a first set of aggregated historical representative risk feature vectors; randomly select a first space point in the set of space points, and construct a first subspace according to a preset space radius; randomly select a space point on the edge of the first subspace as a direction verification space point, and construct a direction verification subspace based on the preset space radius; determine whether the space density of the first subspace is less than or equal to the space density of the direction verification subspace. If so, use the direction from the first space point to the direction verification space point as the iteration direction, and continue to iterate the direction verification space point according to the iteration direction until the preset number of iterations is satisfied, and use the first set of aggregated historical representative risk feature vectors corresponding to the space point obtained in the last iteration as the prototype of the first set of aggregated historical representative risk feature vectors; traverse the K sets of aggregated historical representative risk feature vectors to construct prototypes to obtain the K prototypes of aggregated historical representative risk feature vectors.

[0098] Further, the system is also used for the following functions: When the space density of the first subspace is greater than the space density of the direction verification subspace, re-select the direction verification space point and record the number of re-selections. When the number of re-selections is greater than or equal to the preset number of selections, use the first set of aggregated historical representative risk feature vectors corresponding to the first space point as the prototype of the first set of aggregated historical representative risk feature vectors.

[0099] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. In addition, the above specific embodiments of this specification have been described. Further, the processes depicted in the accompanying drawings do not necessarily require the particular order or sequential order shown to achieve the desired result. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0100] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included within the protection scope of the present application.

[0101] This specification and the drawings are only exemplary descriptions of the present application and are considered to cover any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A method for identifying and positioning risks in a utility tunnel under the co-drive of multiple protocols, characterized in that The method includes: Obtain the pipe gallery structure diagram, deploy multi-source sensors in the pipe gallery based on the pipe gallery structure diagram, and construct a sensor topology network; Use the sensor topology network to continuously monitor the pipe gallery, and transmit the monitoring data to the transmission center of the sensor topology network through multiple communication protocols to generate a real-time sensor monitoring data topology network, where the transmission center includes an intermediate transfer and backtracking database; Based on the real-time sensor monitoring data topology network, perform characterization risk identification, determine characterization risk nodes and characterization risk feature vectors, and perform associated link retrieval in the sensor topology network to determine a target associated link set; For each associated topology node in the target associated link set, extract backtracking data in the intermediate transfer and backtracking database to determine an associated topology node backtracking data sequence set; Perform serialized risk trend identification on the associated topology node backtracking data sequence set to determine a set of implicit risk nodes, and use the characterization risk nodes and the set of implicit risk nodes as the risk identification and positioning results.

2. The method for identifying and positioning risks in a utility tunnel under multi - protocol co - driving according to claim 1, wherein, Obtain the pipe gallery structure diagram, deploy multi-source sensors in the pipe gallery based on the pipe gallery structure diagram, and construct a sensor topology network, including: Extract the pipe gallery route set and the pipe gallery route feature set based on the pipe gallery structure diagram; Perform multi-source sensor layout on the pipe gallery route set respectively according to the pipe gallery route feature set to obtain an initial layout sensor set; Extract the route intersection points in the pipe gallery route set, and deploy multi-source sensors at each route intersection point to obtain an intersection layout sensor set; Perform a union operation on the initial layout sensor set and the intersection layout sensor set to obtain a layout sensor set; Take each layout sensor in the layout sensor set as a topology node, and connect each topology node according to the route direction based on the pipe gallery route set to generate the sensor topology network.

3. The method for identifying and positioning the risks in the utility tunnel under the multi - protocol co - drive as described in claim 1, wherein, Based on the real-time sensor monitoring data topology network, perform characterization risk identification, determine characterization risk nodes and characterization risk feature vectors, including: Obtain the characterization risk identification rule; Use the characterization risk identification rule to perform risk identification on the real-time sensor monitoring data topology network to obtain characterization risk nodes and characterization risk feature vectors.

4. The method for identifying and positioning risks in utility tunnels under multi-protocol parallel drive according to claim 3, wherein Obtain the characterization risk identification rule, including: Obtain the acoustic risk identification rule, where the acoustic risk identification rule is that when the sensor type is a sound sensor, extract the sound frequency and sound amplitude monitored by the sound sensor for anomaly identification. If an anomaly is identified, take the topology node where the sound sensor is located as the characterization risk node, and embed the sound frequency and sound amplitude into an initially empty vector to construct a characterization risk feature vector; Obtaining an optical risk identification rule, wherein the optical risk identification rule is: when the sensor type is an optical sensor, extracting image information monitored by the optical sensor, performing image feature convolution analysis on the image information using a crack image feature identifier to obtain an image feature vector, performing anomaly identification based on the image feature vector, and if an anomaly is identified, using the topological node where the optical sensor is located as a risk node, and using the image feature vector as a risk feature vector; Obtaining a gas risk identification rule, wherein the gas risk identification rule is: when the sensor is a gas sensor, extracting the gas type and gas concentration monitored by the gas sensor for anomaly identification; if an anomaly is identified, using the topological node where the gas sensor is located as a risk representation node, embedding the gas type and gas concentration into an initially empty vector, and constructing a risk representation feature vector; The acoustic risk identification rules, optical risk identification rules and gas risk identification rules are summarized to generate a characterization risk identification rule.

5. A method for identifying and positioning risks in utility tunnels under multi - protocol co - driving according to any one of claims 1 - 4, characterized in that, Performing an associated link search in the sensor topology network to determine a target associated link set includes: Extracting associated pipeline corridor routes from the sensor topology network based on the locations of the risk-representing nodes to obtain an associated pipeline corridor route set; Perform backtracking radius matching based on the risk characteristic vector to obtain an associated backtracking radius; Based on the risk-characterizing nodes and the associated backtracking radius, associated links are extracted from the associated pipeline corridor route set to determine the target associated link set.

6. The method for identifying and positioning risks in a utility tunnel under multi - protocol parallel drive as claimed in claim 5, wherein, Performing backtracking radius matching according to the risk characteristic vector to obtain an associated backtracking radius includes: Pre-built backtracking radius matching table; Performing similarity matching on the risk characterization feature vector and the aggregated historical risk characterization feature vector prototype in the backtracking radius matching table, and taking the aggregated historical risk characterization feature vector prototype corresponding to the maximum similarity as the matching result; The prototype backtracking radius of the matching result is extracted from the backtracking radius matching table as the associated backtracking radius.

7. The method for identifying and positioning risks in a utility tunnel under multi-protocol parallel driving according to claim 6, wherein Pre-built lookback radius matching table, including: Get the historical pipeline corridor risk log collection; Extracting historical correlation distances between historical risk representation nodes and historical implicit risk nodes, and historical risk representation feature vectors in the historical pipeline corridor risk log set to obtain a historical risk representation feature vector set and a historical correlation distance set; Aggregating the historical risk characterization feature vector sets by the same type to obtain K aggregated historical risk characterization feature vector sets, where K is a positive integer; Mapping and aggregating the historical correlation distance set according to the K aggregated historical risk characterization feature vector sets to obtain K aggregated historical correlation distance sets; Prototype the K aggregated historical risk feature vector sets to generate K aggregated historical risk feature vector prototypes; The maximum value in the K aggregated historical association distance sets is used as the K prototype backtracking radius; A backtracking radius matching table is constructed according to the K aggregated historical risk feature vector prototypes and the K prototype backtracking radii.

8. The method for identifying and positioning risks in a utility tunnel under multi-protocol parallel driving according to claim 7, wherein Perform prototype construction on the set of K aggregated historical representation risk feature vectors to generate K prototypes of aggregated historical representation risk feature vectors, including: Extract the first set of aggregated historical representation risk feature vectors from the set of K aggregated historical representation risk feature vectors; Map the first set of aggregated historical representation risk feature vectors to an M-dimensional space respectively to obtain a set of space points, where M is a positive integer, and the coordinates of each space point correspond to a first aggregated historical representation risk feature vector; Randomly select a first space point from the set of space points and construct a first subspace according to a preset space radius; Randomly select a space point on the edge of the first subspace as a direction verification space point, and construct a direction verification subspace based on the preset space radius; Judge whether the space density of the first subspace is less than or equal to the space density of the direction verification subspace. If so, use the direction from the first space point to the direction verification space point as the iteration direction, and continue to iterate the direction verification space point in accordance with the iteration direction until the preset number of iterations is satisfied. Use the first aggregated historical representation risk feature vector corresponding to the space point obtained in the last iteration as the prototype of the first aggregated historical representation risk feature vector; Traverse the set of K aggregated historical representation risk feature vectors to perform prototype construction to obtain the K prototypes of aggregated historical representation risk feature vectors.

9. The method for identifying and positioning risks in a utility tunnel under multi-protocol co-driving according to claim 8, wherein When the space density of the first subspace is greater than the space density of the direction verification subspace, reselect the direction verification space point and record the number of reselections. When the number of reselections is greater than or equal to the preset number of selections, use the first aggregated historical representation risk feature vector corresponding to the first space point as the prototype of the first aggregated historical representation risk feature vector.

10. A risk identification and positioning system for utility tunnels under multi - protocol co - driving, characterized in that, The system is used to implement the method for identifying and locating risks in a pipe gallery under the drive of multiple protocols according to any one of claims 1-9. The system includes: A sensor topology network construction module, configured to obtain a pipe gallery structure diagram, deploy multi-source sensors in the pipe gallery based on the pipe gallery structure diagram, and construct a sensor topology network; A topology network generation module, configured to use the sensor topology network to continuously monitor the pipe gallery, and transmit monitoring data to the transmission center of the sensor topology network through multiple communication protocols to generate a real-time sensor monitoring data topology network, where the transmission center includes an intermediate transfer and backtracking database; An associated link set determination module, configured to perform characterization risk identification based on the real-time sensor monitoring data topology network, determine characterization risk nodes and characterization risk feature vectors, and perform associated link retrieval in the sensor topology network to determine a target associated link set; A backtracking data extraction module, configured to extract backtracking data in the intermediate transfer and backtracking database for each associated topology node in the target associated link set to determine a set of associated topology node backtracking data sequences; An implicit risk node set determination module, which is used to perform serialized risk trend identification on the backtracking data sequence set of the associated topological nodes, determine the implicit risk node set, and use the represented risk nodes and the implicit risk node set as the risk identification and positioning results.

Images