A risk early warning method and system for underground utility tunnels

Abstract

This invention discloses a risk early warning method and system for underground utility tunnel spaces, applied in the field of underground utility tunnel engineering technology. The method includes: acquiring historical sensor monitoring data and geospatial data of the underground utility tunnel space; constructing a spatiotemporal risk element map; building a risk assessment and disposal knowledge base based on historical experience data and standard contingency plan data; extracting location nodes and sensor connectivity directions associated with abnormal events from the spatiotemporal risk element map to form a spatiotemporal anomaly local element map; obtaining an assessment and disposal knowledge set based on the abnormal events; inputting the spatiotemporal anomaly local element map and the assessment and disposal knowledge set into an anomaly detection model for processing to obtain an anomaly factor set; and processing the anomaly factor set and the spatiotemporal anomaly local element map to generate a risk early warning scheme. The method provided by this invention can solve the technical problems of isolated early warning information and one-sided risk assessment in existing risk early warning methods.

Description

Technical Field

[0001] This invention relates to the field of underground utility tunnel engineering technology, and in particular to a risk warning method and system for underground utility tunnel spaces. Background Technology

[0002] Power utility tunnels are characterized by enclosed spaces, dense facilities, coupled pipelines, continuous operation, and significant external environmental disturbances. The spatial risks of utility tunnels not only come from internal monitoring anomalies such as water level, temperature and humidity, and equipment failures, but are also affected by external spatial factors such as population distribution and construction disturbances.

[0003] In existing technologies, risk warning methods are generally based on sensor threshold monitoring and alarm methods. This method uses the exceeding of a single indicator limit as the trigger condition, which cannot unify the spatiotemporal aggregation of abnormal events with multiple influencing factors, nor can it explain the possible impact range, affected nodes and propagation lines of abnormal events in the utility tunnel space. This easily leads to isolated warning information and one-sided risk assessment, thus failing to achieve accurate risk warning and accurate processing of risk warnings. Summary of the Invention

[0004] This invention provides a risk warning method and system for underground utility tunnels, which can solve the technical problems of isolated warning information and one-sided risk assessment in existing risk warning methods, and realize accurate warning and accurate handling of risks in underground utility tunnels.

[0005] To address the aforementioned technical problems, this invention provides a risk warning method for underground utility tunnel spaces, the method comprising: Acquire historical sensor monitoring data and geospatial data of underground utility tunnel space; A spatiotemporal risk element map is constructed based on the historical sensor monitoring data and the geospatial data, and a risk assessment and disposal knowledge base is constructed based on the acquired historical experience data and standard contingency plan data. Based on the abnormal events in the underground utility tunnel space monitored in real time, the positioning nodes and sensing connection directions associated with the abnormal events are extracted from the spatiotemporal risk element map, and a spatiotemporal anomaly local element map is formed based on the positioning nodes and the sensing connection directions. Based on the abnormal event, recall knowledge associated with the abnormal event is extracted from the risk assessment and disposal knowledge base, and the recall knowledge is verified and prioritized to obtain a set of assessment and disposal knowledge. The spatiotemporal anomaly local element map and the judgment and disposal knowledge set are input into the anomaly detection model constructed by the time series prediction algorithm for processing to obtain the anomaly factor set; The set of anomalous factors and the spatiotemporal anomalous local feature map are processed, and a risk warning scheme for the underground utility tunnel space is generated based on the determined anomalous increment map.

[0006] As a preferred embodiment, before constructing the spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, the method further includes: The historical sensor monitoring data and the geospatial data are cleaned and standardized, and spatial and temporal references are unified. The data cleaning and standardization includes field mapping, unit unification, missing value handling, anomaly screening, and duplicate record elimination; the spatial benchmark unification includes coordinate transformation, geocoding, buffer calculation, spatial overlay, and topological relationship construction; the temporal benchmark unification includes timestamp correction, sampling frequency alignment, time window division, periodic feature extraction, and event duration calculation.

[0007] As one preferred embodiment, the spatiotemporal anomaly local feature map and the judgment and determination knowledge set are input into an anomaly detection model constructed by a time-series prediction algorithm for processing, to obtain an anomaly factor set, including: The attribute vector of the local node in the spatiotemporal anomaly local feature map, the connection weight vector of the sensing connectivity direction, and the time series fluctuation vector along the sensing connectivity direction are obtained and spliced ​​together to form a first multidimensional feature tensor. The recalled knowledge in the judgment and disposal knowledge set is semantically encoded, and knowledge fragments related to the current abnormal event in the judgment and disposal knowledge set are aggregated through an attention mechanism to generate a second semantic feature vector. The first multidimensional feature tensor and the second semantic feature vector are aligned and fused along the channel dimension to obtain the fused feature representation; The fused feature representation is input into the anomaly detection model constructed by the time-series prediction algorithm. After multi-layer iterative processing, the anomaly probability sequence and anomaly intensity sequence are output. The abnormal probability sequence and the abnormal intensity sequence are sequentially subjected to confidence threshold filtering and non-maximum suppression to obtain the abnormal factor set.

[0008] As one preferred embodiment, the risk warning scheme for generating the underground utility tunnel space based on a determined anomaly increment map includes: Based on the risk prediction increment in the anomaly increment map and the preset line division rule template in the risk assessment and disposal knowledge base, a comprehensive line division assessment is performed to generate the risk warning scheme; The risk warning plan should at least include the risk classification, level, scope of impact, temporal evolution trend, standard basis, and disposal suggestions for the underground utility tunnel space.

[0009] As one preferred embodiment, after obtaining the aforementioned risk warning scheme, the risk warning method for underground utility tunnel space includes: Based on the proposed solutions, a standardized emergency response work order is generated and dispatched to the corresponding maintenance team to initiate the response process and track the execution status, progress, and results of the response measures.

[0010] The present invention also provides a risk warning system for underground utility tunnel spaces, comprising: The acquisition module is used to acquire historical sensor monitoring data and geospatial data of the underground utility tunnel space; The module is used to construct a spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, and to construct a risk assessment and disposal knowledge base based on the acquired historical experience data and standard contingency plan data. An anomaly local extraction module is used to extract the location nodes and sensor connection directions associated with the anomalies in the underground utility tunnel space as monitored in real time from the spatiotemporal risk element map, and to form a spatiotemporal anomaly local element map based on the location nodes and the sensor connection directions. The risk assessment and disposal extraction module is used to extract recall knowledge associated with the abnormal event from the risk assessment and disposal knowledge base based on the abnormal event, and to perform verification and priority sorting on the recall knowledge to obtain a set of assessment and disposal knowledge. The processing module is used to input the spatiotemporal anomaly local element map and the judgment and disposal knowledge set into the anomaly detection model constructed by the time series prediction algorithm for processing, so as to obtain the anomaly factor set; The generation module is used to process the set of abnormal factors and the spatiotemporal abnormal local feature map, and generate a risk warning scheme for the underground utility tunnel space based on the determined abnormal incremental map.

[0011] As a preferred embodiment, before constructing the spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, the system further includes: The historical sensor monitoring data and the geospatial data are cleaned and standardized, and spatial and temporal references are unified. The data cleaning and standardization includes field mapping, unit unification, missing value handling, anomaly screening, and duplicate record elimination; the spatial benchmark unification includes coordinate transformation, geocoding, buffer calculation, spatial overlay, and topological relationship construction; the temporal benchmark unification includes timestamp correction, sampling frequency alignment, time window division, periodic feature extraction, and event duration calculation.

[0012] As one preferred embodiment, the spatiotemporal anomaly local feature map and the judgment and determination knowledge set are input into an anomaly detection model constructed by a time-series prediction algorithm for processing, to obtain an anomaly factor set, including: The attribute vector of the local node in the spatiotemporal anomaly local feature map, the connection weight vector of the sensing connectivity direction, and the time series fluctuation vector along the sensing connectivity direction are obtained and spliced ​​together to form a first multidimensional feature tensor. The recalled knowledge in the judgment and disposal knowledge set is semantically encoded, and knowledge fragments related to the current abnormal event in the judgment and disposal knowledge set are aggregated through an attention mechanism to generate a second semantic feature vector. The first multidimensional feature tensor and the second semantic feature vector are aligned and fused along the channel dimension to obtain the fused feature representation; The fused feature representation is input into the anomaly detection model constructed by the time-series prediction algorithm. After multi-layer iterative processing, the anomaly probability sequence and anomaly intensity sequence are output. The abnormal probability sequence and the abnormal intensity sequence are sequentially subjected to confidence threshold filtering and non-maximum suppression to obtain the abnormal factor set.

[0013] As one preferred embodiment, the risk warning scheme for generating the underground utility tunnel space based on a determined anomaly increment map includes: Based on the risk prediction increment in the anomaly increment map and the preset line division rule template in the risk assessment and disposal knowledge base, a comprehensive line division assessment is performed to generate the risk warning scheme; The risk warning plan should at least include the risk classification, level, scope of impact, temporal evolution trend, standard basis, and disposal suggestions for the underground utility tunnel space.

[0014] As one preferred embodiment, after obtaining the aforementioned risk warning scheme, the risk warning system for underground utility tunnel space includes: Based on the proposed solutions, a standardized emergency response work order is generated and dispatched to the corresponding maintenance team to initiate the response process and track the execution status, progress, and results of the response measures.

[0015] Compared with the prior art, the beneficial effects of the present invention are at least one of the following: This invention acquires historical sensor monitoring data and geospatial data of underground utility tunnel space; constructs a spatiotemporal risk element map based on the historical sensor monitoring data and geospatial data, and constructs a risk assessment and disposal knowledge base based on the acquired historical experience data and standard contingency plan data; extracts location nodes and sensor connection directions associated with the abnormal events from the spatiotemporal risk element map based on the real-time monitored abnormal events of the underground utility tunnel space, and forms a spatiotemporal abnormal local element map based on the location nodes and sensor connection directions; extracts recall knowledge associated with the abnormal events from the risk assessment and disposal knowledge base based on the abnormal events, verifies and prioritizes the recalled knowledge to obtain an assessment and disposal knowledge set; inputs the spatiotemporal abnormal local element map and the assessment and disposal knowledge set into an anomaly detection model constructed by a time-series prediction algorithm for processing to obtain an anomaly factor set; processes the anomaly factor set and the spatiotemporal abnormal local element map, and generates a risk warning scheme for the underground utility tunnel space based on the determined anomaly increment map.

[0016] Compared with existing technologies, this invention can dynamically associate multi-source spatiotemporal elements to form a unified spatiotemporal anomaly local map; combine empirical knowledge and standard contingency plans to generate a verified and sorted set of judgment and disposal knowledge; output anomaly factor set through anomaly detection model, and generate a feasible risk warning plan based on the anomaly incremental map, thereby improving the completeness of the warning and the accuracy of the warning for the risks of underground utility tunnel space. Attached Figure Description

[0017] Figure 1 This is a flowchart illustrating a risk warning method for underground utility tunnel space according to one embodiment of the present invention. Figure 2 This is a schematic diagram of the structure of a risk warning system for underground utility tunnel space according to one embodiment of the present invention; Figure label: The module consists of: 11. Acquisition module; 12. Construction module; 13. Anomaly local extraction module; 14. Analysis and handling extraction module; 15. Processing module; and 16. Generation module. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. The purpose of providing these embodiments is to make the disclosure of the present invention more thorough and comprehensive. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative effort are within the scope of protection of the present invention.

[0019] In the description of this invention, the terms "first," "second," "third," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, a feature defined with "first," "second," "third," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, unless otherwise stated, "a plurality of" means two or more.

[0020] In the description of this invention, it should be noted that, unless otherwise defined, all technical and scientific terms used in this invention have the same meaning as commonly understood by one of ordinary skill in the art. The terminology used in this specification is for the purpose of describing specific embodiments only and is not intended to limit the invention. Those skilled in the art can understand the specific meaning of the above terms in this invention based on the specific circumstances.

[0021] One embodiment of the present invention provides a risk warning method for underground utility tunnel spaces. For details, please refer to [link / reference]. Figure 1 , Figure 1 The diagram shown is a flowchart illustrating a risk warning method for underground utility tunnel spaces according to one embodiment of the present invention. The method includes: S1: Acquire historical sensor monitoring data and geospatial data of the underground utility tunnel space; S2: Construct a spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, and construct a risk assessment and disposal knowledge base based on the acquired historical experience data and standard contingency plan data; S3: Based on the abnormal events in the underground utility tunnel space monitored in real time, extract the positioning nodes and sensing connection directions associated with the abnormal events from the spatiotemporal risk element map, and form a spatiotemporal abnormal local element map based on the positioning nodes and the sensing connection directions. S4: Based on the abnormal event, extract recall knowledge associated with the abnormal event from the risk assessment and disposal knowledge base, perform verification and priority sorting on the recall knowledge, and obtain a set of assessment and disposal knowledge. S5: Input the spatiotemporal anomaly local element map and the judgment and disposal knowledge set into the anomaly detection model constructed by the time series prediction algorithm for processing to obtain the anomaly factor set; S6: Process the set of abnormal factors and the spatiotemporal abnormal local element map, and generate a risk warning scheme for the underground utility tunnel space based on the determined abnormal incremental map.

[0022] Specifically, historical sensor monitoring data and geospatial data of the underground utility tunnel space are acquired. The historical sensor monitoring data is raw factual data obtained directly through monitoring, business records, and spatial acquisition, such as meteorological data, historical accidents, population distribution, remote sensing data, and sensor data. It is the basic raw information supporting scene analysis. The geospatial data is POI geospatial data, which provides spatial location reference for the scene.

[0023] The historical sensor monitoring data and the geospatial data are subjected to data cleaning and standardization, spatial benchmark unification, and temporal benchmark unification. The data cleaning and standardization includes field mapping, unit unification, missing value handling, anomaly screening, and duplicate record elimination. The spatial benchmark unification includes coordinate transformation, geocoding, buffer calculation, spatial overlay, and topological relationship construction. The temporal benchmark unification includes timestamp correction, sampling frequency alignment, time window division, periodic feature extraction, and event duration calculation.

[0024] Based on the preprocessed historical sensor monitoring data and the geospatial data, a spatiotemporal risk element map is constructed. Specifically, entities such as pipe gallery sections, manholes, equipment, sensors, and monitoring indicators are extracted from the historical sensor monitoring data, and entities such as meteorological areas, remote sensing anomaly patches, historical accident points, points of interest, population exposure areas, and roads and rivers are extracted from the geospatial data. Each entity node is assigned a set of attributes including spatial coordinates, time validity, and type labels to obtain a node set.

[0025] Based on the spatial coordinates of each node in the node set and the pipe gallery design drawings, spatial proximity edges, topological connectivity edges, equipment affiliation edges, drainage connectivity edges, ventilation connectivity edges, and cable connection edges are established. Indicator correlation edges and time overlap edges are established based on the time series correlation of the historical sensor monitoring data. The spatiotemporal risk element map is obtained based on the node set and all edges.

[0026] At the same time, a risk assessment and disposal knowledge base is built based on standards and specifications, historical accidents, expert experience, handling records, and operation and maintenance plans.

[0027] Based on the abnormal events in the underground utility tunnel space monitored in real time, the positioning nodes and sensor connection directions associated with the abnormal events are extracted from the spatiotemporal risk element map. A spatiotemporal anomaly local element map is formed based on the positioning nodes and sensor connection directions. The abnormal events are generated by sensor exceeding limits, abnormal trends, video alarms, equipment failures, external environmental events, manual reporting, or standard specification triggering conditions, and at least include the abnormal object, abnormal index, abnormal value, occurrence time, spatial location, section, candidate abnormal type, initial level, and confidence level.

[0028] Therefore, when the system detects monitoring limits exceeding limits, abnormal trends, video alarms, equipment failures, external environmental events, or manual reports, it generates abnormal events. Based on the abnormal object, abnormal indicators, occurrence time, spatial location, and abnormal type candidate of the abnormal event, the corresponding node or spatial range is located in the spatiotemporal risk element map. The abnormal event is then coupled with relevant sensors, equipment, sections, environmental objects, and historical accident points to form anchor points for subsequent local map extraction.

[0029] Then, using the anomaly anchor point as the center, the spatiotemporal risk element map is segmented according to preset spatial radius, time window, topological hop count, and physical connectivity direction to generate a spatiotemporal anomaly local element map. The spatiotemporal anomaly local element map includes the anomaly source node, adjacent segments, connectivity edges, sensor indicators, meteorological remote sensing triggers, historical accident points, sensitive targets, and exposed population objects related to the current anomaly.

[0030] It should be noted that the spatiotemporal anomaly local element map only retains local facts related to the current anomaly assessment, to avoid noise caused by including the full map in subsequent predictions.

[0031] Based on the abnormal event, recall knowledge associated with the abnormal event is extracted from the risk assessment and disposal knowledge base. The recall knowledge is then verified and prioritized to obtain a set of assessment and disposal knowledge.

[0032] Specifically, using the anomaly type, anomaly object, spatial conditions, environmental causes, similarity of historical accidents, and sensitive target exposure information in the spatiotemporal anomaly local element map as search criteria, relevant standards and specifications, historical cases, risk scenario rules, disposal measures, predictive factor suggestions, and branch conclusion templates are retrieved from the risk assessment and disposal knowledge base.

[0033] The recalled knowledge is sequentially subjected to applicability verification, conflict resolution, and priority sorting to generate a set of judgment and disposal knowledge based on the local anomaly map.

[0034] The spatiotemporal anomaly local feature map and the assessment and judgment knowledge set are input into an anomaly detection model constructed by a time-series prediction algorithm for processing to obtain an anomaly factor set. This includes: acquiring the attribute vectors of the localized nodes in the spatiotemporal anomaly local feature map, the connection weight vectors of the sensing connectivity direction, and the time-series fluctuation vectors along the sensing connectivity direction, and concatenating them to form a first multidimensional feature tensor; semantically encoding the recalled knowledge in the assessment and judgment knowledge set, and aggregating knowledge fragments related to the current anomaly event within the assessment and judgment knowledge set through an attention mechanism to generate a second semantic feature vector; aligning and fusing the first multidimensional feature tensor and the second semantic feature vector in the channel dimension to obtain a fused feature representation; inputting the fused feature representation into the anomaly detection model constructed by the time-series prediction algorithm, and outputting an anomaly probability sequence and an anomaly intensity sequence after multi-layer iterative processing; and sequentially performing confidence threshold filtering and non-maximum suppression processing on the anomaly probability sequence and the anomaly intensity sequence to obtain the anomaly factor set.

[0035] The time series prediction algorithm can be implemented using ARIMA, TCN, GRU, LSTM, Transformer, Temporal FusionTransformer, PatchTST, graph time series network or a combination thereof.

[0036] The set of anomalous factors includes the predicted values ​​of anomalous factors, node risk probabilities, risk upward trends, anomalous duration, risk arrival time, and prediction confidence levels for multiple future time slices.

[0037] In another embodiment, the spatiotemporal anomaly local feature map and the judgment and disposal knowledge set are input into the anomaly detection model constructed by the time series prediction algorithm for processing to obtain an anomaly factor set. The method also includes determining predictable factors and prediction nodes based on the spatiotemporal anomaly local feature map and the anomaly scene knowledge set, and predicting the predictable factors and prediction nodes through the time series prediction algorithm to obtain the anomaly factor set.

[0038] The set of anomalous factors and the spatiotemporal anomalous local element map are processed to generate a risk warning scheme for the underground utility tunnel space based on the determined anomalous increment map. Specifically, based on the risk prediction increment in the anomalous increment map and the preset delineation rule template in the risk assessment and disposal knowledge base, a comprehensive delineation assessment is performed to generate the risk warning scheme. The risk warning scheme includes at least the risk delineation, level, impact range, temporal evolution trend, standard basis, and disposal suggestions for the underground utility tunnel space.

[0039] It should also be noted that the risk warning scheme outputs risk analysis results based on the abnormal incremental map as the predictive basis and the risk assessment and disposal knowledge base as the knowledge basis. These results include risk level, risk sub-line, scope of impact, temporal evolution trend, standard basis, disposal suggestions, warning targets, and confidence level of conclusions.

[0040] After receiving the aforementioned risk warning plan, a standardized emergency response work order is generated based on the handling suggestions and dispatched to the corresponding operation and maintenance team to initiate the handling process and track the execution status, progress and results of the handling measures.

[0041] The specific implementation method is as follows: A certain underground utility tunnel area experienced continuous heavy rainfall. Meteorological data, remote sensing data on water accumulation, and water level sensor data for section A of the underground utility tunnel area were acquired. After time alignment, spatial registration, and unit unification of the above data, a dataset was formed. The water level in section A rose from 15 cm to 42 cm within 10 minutes. Remote sensing identified water accumulation patches on adjacent roads, and meteorological data indicated that heavy rainfall was expected to continue for the next hour.

[0042] Based on the dataset, a spatiotemporal risk element map is constructed using the pipe gallery sections, drainage paths, water level sensors, road flooding patches, historical flooding incident sites, and surrounding sensitive targets. Simultaneously, the system stores flooding backflow standard clauses, historical flooding cases, drainage pump handling experience, connecting section sealing measures, and flood control response level rules in the risk assessment and disposal knowledge base.

[0043] An abnormal event is generated when the water level in section A exceeds the warning threshold.

[0044] The system couples abnormal events with the water level sensor in section A, section A, the AB connecting section, drainage pumps, road flooding patches, and historical accident points in the spatiotemporal risk element map to form a spatiotemporal anomaly local element map. The system then retrieves knowledge items such as "continuous heavy rainfall—road flooding—manhole water ingress—section water level rise—connecting section backflow" from the risk assessment and disposal knowledge base, along with corresponding standard clauses and disposal rules, to form an assessment and disposal knowledge set.

[0045] The temporal prediction algorithm predicts the spatiotemporal anomaly local element map and the judgment and disposal knowledge set to form an anomaly factor set.

[0046] The set of anomalous factors indicates a high probability of continued water level rise in section A within the next 40 minutes, a risk of backflow in the AB connecting section, and an increased probability of section B being affected. These predictions are written into the spatiotemporal anomaly local feature map to generate an anomaly increment map, which adds a water level prediction node for section A, a water inflow probability node for section B, a predicted propagation edge for the AB connecting section, and a 60-minute response window.

[0047] The system combines the abnormal incremental map with the risk assessment and disposal knowledge base to output a risk warning plan: backflow of waterlogging is a high risk, and the main impact path is "heavy rainfall - road water accumulation - water level rise in section A - backflow in section AB - section B affected"; it is recommended to start drainage pump verification, emergency pumping, inspection and sealing of section AB, and advance inspection of section B within 60 minutes; if the water level continues to rise after 30 minutes, the response level should be upgraded.

[0048] Another embodiment of the present invention provides a risk early warning system for underground utility tunnel spaces. For details, please refer to [link to relevant documentation]. Figure 2 , Figure 2 The diagram shown is a structural schematic of a risk warning system for underground utility tunnel spaces according to one embodiment of the present invention. The system includes: Module 11 is used to acquire historical sensor monitoring data and geospatial data of the underground utility tunnel space; Module 12 is used to construct a spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, and to construct a risk assessment and disposal knowledge base based on the acquired historical experience data and standard contingency plan data. The abnormal local extraction module 13 is used to extract the positioning nodes and sensing connection directions associated with the abnormal events from the spatiotemporal risk element map based on the abnormal events of the underground utility tunnel space monitored in real time, and to form a spatiotemporal abnormal local element map based on the positioning nodes and the sensing connection directions. The risk assessment and disposal extraction module 14 is used to extract recall knowledge associated with the abnormal event from the risk assessment and disposal knowledge base based on the abnormal event, and to perform verification and priority sorting on the recall knowledge to obtain a set of assessment and disposal knowledge. Processing module 15 is used to input the spatiotemporal anomaly local element map and the judgment and disposal knowledge set into the anomaly detection model constructed by the time series prediction algorithm for processing, so as to obtain an anomaly factor set; The generation module 16 is used to process the set of abnormal factors and the spatiotemporal abnormal local feature map, and generate a risk warning scheme for the underground utility tunnel space based on the determined abnormal incremental map.

[0049] As a preferred embodiment, before constructing the spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, the system further includes: The historical sensor monitoring data and the geospatial data are cleaned and standardized, and spatial and temporal references are unified. The data cleaning and standardization includes field mapping, unit unification, missing value handling, anomaly screening, and duplicate record elimination; the spatial benchmark unification includes coordinate transformation, geocoding, buffer calculation, spatial overlay, and topological relationship construction; the temporal benchmark unification includes timestamp correction, sampling frequency alignment, time window division, periodic feature extraction, and event duration calculation.

[0050] As one preferred embodiment, the spatiotemporal anomaly local feature map and the judgment and determination knowledge set are input into an anomaly detection model constructed by a time-series prediction algorithm for processing, to obtain an anomaly factor set, including: The attribute vector of the local node in the spatiotemporal anomaly local feature map, the connection weight vector of the sensing connectivity direction, and the time series fluctuation vector along the sensing connectivity direction are obtained and spliced ​​together to form a first multidimensional feature tensor. The recalled knowledge in the judgment and disposal knowledge set is semantically encoded, and knowledge fragments related to the current abnormal event in the judgment and disposal knowledge set are aggregated through an attention mechanism to generate a second semantic feature vector. The first multidimensional feature tensor and the second semantic feature vector are aligned and fused along the channel dimension to obtain the fused feature representation; The fused feature representation is input into the anomaly detection model constructed by the time-series prediction algorithm. After multi-layer iterative processing, the anomaly probability sequence and anomaly intensity sequence are output. The abnormal probability sequence and the abnormal intensity sequence are sequentially subjected to confidence threshold filtering and non-maximum suppression to obtain the abnormal factor set.

[0051] As one preferred embodiment, the risk warning scheme for generating the underground utility tunnel space based on a determined anomaly increment map includes: Based on the risk prediction increment in the anomaly increment map and the preset line division rule template in the risk assessment and disposal knowledge base, a comprehensive line division assessment is performed to generate the risk warning scheme; The risk warning plan should at least include the risk classification, level, scope of impact, temporal evolution trend, standard basis, and disposal suggestions for the underground utility tunnel space.

[0052] As one preferred embodiment, after obtaining the aforementioned risk warning scheme, the risk warning system for underground utility tunnel space includes: Based on the proposed solutions, a standardized emergency response work order is generated and dispatched to the corresponding maintenance team to initiate the response process and track the execution status, progress, and results of the response measures.

[0053] The embodiments described above are merely illustrative of several implementations of the present invention, and while the descriptions are specific and detailed, they should not be construed as limiting the scope of the present invention. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of the present invention, and these modifications and improvements all fall within the scope of protection of the present invention. Therefore, the scope of protection of this patent should be determined by the appended claims.

Claims

1. A risk early warning method for underground utility tunnel spaces, characterized in that, include: Acquire historical sensor monitoring data and geospatial data of underground utility tunnel space; A spatiotemporal risk element map is constructed based on the historical sensor monitoring data and the geospatial data, and a risk assessment and disposal knowledge base is constructed based on the acquired historical experience data and standard contingency plan data. Based on the abnormal events in the underground utility tunnel space monitored in real time, the positioning nodes and sensing connection directions associated with the abnormal events are extracted from the spatiotemporal risk element map, and a spatiotemporal anomaly local element map is formed based on the positioning nodes and the sensing connection directions. Based on the abnormal event, recall knowledge associated with the abnormal event is extracted from the risk assessment and disposal knowledge base, and the recall knowledge is verified and prioritized to obtain a set of assessment and disposal knowledge. The spatiotemporal anomaly local element map and the judgment and disposal knowledge set are input into the anomaly detection model constructed by the time series prediction algorithm for processing to obtain the anomaly factor set; The set of anomalous factors and the spatiotemporal anomalous local feature map are processed, and a risk warning scheme for the underground utility tunnel space is generated based on the determined anomalous increment map.

2. The risk early warning method for underground utility tunnel space as described in claim 1, characterized in that, Before constructing a spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, the method further includes: The historical sensor monitoring data and the geospatial data are cleaned and standardized, and spatial and temporal references are unified. The data cleaning and standardization includes field mapping, unit unification, missing value handling, anomaly screening, and duplicate record elimination; the spatial benchmark unification includes coordinate transformation, geocoding, buffer calculation, spatial overlay, and topological relationship construction; the temporal benchmark unification includes timestamp correction, sampling frequency alignment, time window division, periodic feature extraction, and event duration calculation.

3. The risk early warning method for underground utility tunnel space as described in claim 1, characterized in that, The process involves inputting the spatiotemporal anomaly local feature map and the judgment and determination knowledge set into an anomaly detection model constructed by a time-series prediction algorithm for processing, resulting in an anomaly factor set, including: The attribute vector of the local node in the spatiotemporal anomaly local feature map, the connection weight vector of the sensing connectivity direction, and the time series fluctuation vector along the sensing connectivity direction are obtained and spliced ​​together to form a first multidimensional feature tensor. The recalled knowledge in the judgment and disposal knowledge set is semantically encoded, and knowledge fragments related to the current abnormal event in the judgment and disposal knowledge set are aggregated through an attention mechanism to generate a second semantic feature vector. The first multidimensional feature tensor and the second semantic feature vector are aligned and fused along the channel dimension to obtain the fused feature representation; The fused feature representation is input into the anomaly detection model constructed by the time-series prediction algorithm. After multi-layer iterative processing, the anomaly probability sequence and anomaly intensity sequence are output. The abnormal probability sequence and the abnormal intensity sequence are sequentially subjected to confidence threshold filtering and non-maximum suppression to obtain the abnormal factor set.

4. The risk early warning method for underground utility tunnel space as described in claim 1, characterized in that, The risk warning scheme for generating the underground utility tunnel space based on the determined anomaly increment map includes: Based on the risk prediction increment in the anomaly increment map and the preset line division rule template in the risk assessment and disposal knowledge base, a comprehensive line division assessment is performed to generate the risk warning scheme; The risk warning plan should at least include the risk classification, level, scope of impact, temporal evolution trend, standard basis, and disposal suggestions for the underground utility tunnel space.

5. The risk early warning method for underground utility tunnel space as described in claim 4, characterized in that, After obtaining the aforementioned risk warning scheme, the risk warning method for underground utility tunnel spaces includes: Based on the proposed solutions, a standardized emergency response work order is generated and dispatched to the corresponding maintenance team to initiate the response process and track the execution status, progress, and results of the response measures.

6. A risk early warning system for underground utility tunnel spaces, characterized in that, include: The acquisition module is used to acquire historical sensor monitoring data and geospatial data of the underground utility tunnel space; The module is used to construct a spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, and to construct a risk assessment and disposal knowledge base based on the acquired historical experience data and standard contingency plan data. An anomaly local extraction module is used to extract the location nodes and sensor connection directions associated with the anomalies in the underground utility tunnel space as monitored in real time from the spatiotemporal risk element map, and to form a spatiotemporal anomaly local element map based on the location nodes and the sensor connection directions. The risk assessment and disposal extraction module is used to extract recall knowledge associated with the abnormal event from the risk assessment and disposal knowledge base based on the abnormal event, and to perform verification and priority sorting on the recall knowledge to obtain a set of assessment and disposal knowledge. The processing module is used to input the spatiotemporal anomaly local element map and the judgment and disposal knowledge set into the anomaly detection model constructed by the time series prediction algorithm for processing, so as to obtain the anomaly factor set; The generation module is used to process the set of abnormal factors and the spatiotemporal abnormal local feature map, and generate a risk warning scheme for the underground utility tunnel space based on the determined abnormal incremental map.

7. The risk early warning system for underground utility tunnels as described in claim 6, characterized in that, Before constructing a spatiotemporal risk element map based on the historical sensor monitoring data and the geospatial data, the system further includes: The historical sensor monitoring data and the geospatial data are cleaned and standardized, and spatial and temporal references are unified. The data cleaning and standardization includes field mapping, unit unification, missing value handling, anomaly screening, and duplicate record elimination; the spatial benchmark unification includes coordinate transformation, geocoding, buffer calculation, spatial overlay, and topological relationship construction; the temporal benchmark unification includes timestamp correction, sampling frequency alignment, time window division, periodic feature extraction, and event duration calculation.

8. The risk early warning system for underground utility tunnel spaces as described in claim 6, characterized in that, The process involves inputting the spatiotemporal anomaly local feature map and the judgment and determination knowledge set into an anomaly detection model constructed by a time-series prediction algorithm for processing, resulting in an anomaly factor set, including: The attribute vector of the local node in the spatiotemporal anomaly local feature map, the connection weight vector of the sensing connectivity direction, and the time series fluctuation vector along the sensing connectivity direction are obtained and spliced ​​together to form a first multidimensional feature tensor. The recalled knowledge in the judgment and disposal knowledge set is semantically encoded, and knowledge fragments related to the current abnormal event in the judgment and disposal knowledge set are aggregated through an attention mechanism to generate a second semantic feature vector. The first multidimensional feature tensor and the second semantic feature vector are aligned and fused along the channel dimension to obtain the fused feature representation; The fused feature representation is input into the anomaly detection model constructed by the time-series prediction algorithm. After multi-layer iterative processing, the anomaly probability sequence and anomaly intensity sequence are output. The abnormal probability sequence and the abnormal intensity sequence are sequentially subjected to confidence threshold filtering and non-maximum suppression to obtain the abnormal factor set.

9. The risk early warning system for underground utility tunnel spaces as described in claim 6, characterized in that, The risk warning scheme for generating the underground utility tunnel space based on the determined anomaly increment map includes: Based on the risk prediction increment in the anomaly increment map and the preset line division rule template in the risk assessment and disposal knowledge base, a comprehensive line division assessment is performed to generate the risk warning scheme; The risk warning plan should at least include the risk classification, level, scope of impact, temporal evolution trend, standard basis, and disposal suggestions for the underground utility tunnel space.

10. The risk early warning system for underground utility tunnel spaces as described in claim 9, characterized in that, After obtaining the aforementioned risk warning scheme, the risk warning system for underground utility tunnel spaces includes: Based on the proposed solutions, a standardized emergency response work order is generated and dispatched to the corresponding maintenance team to initiate the response process and track the execution status, progress, and results of the response measures.

Images