Deep foundation pit multi-parameter intelligent monitoring method and system based on Internet of Things, medium and product

Abstract

The invention provides a deep foundation pit multi-parameter intelligent monitoring method and system based on the Internet of Things, a medium and a product, and relates to the technical field of deep foundation pit monitoring, and the method comprises the steps: obtaining multi-type monitoring data of a deep foundation pit, determining a cross-physical-quantity conduction relationship among all monitoring parameters in the data, and constructing a heterogeneous association graph structure according to the relationship; performing feature extraction on the data on the graph structure to obtain a time sequence change feature and a conduction influence feature, and performing dynamic weight disturbance operation on an associated edge according to the two features to obtain a weight stability index; marking the associated edge with the index greater than a preset stable threshold value in the graph structure as an effective conduction edge, and determining a stable conduction path set according to the connectivity of the effective conduction edge in the graph structure; and when the target conduction path meeting the collaborative risk triggering condition exists in the set, determining that the monitoring node in the path is in a multi-parameter collaborative risk state, and executing risk response intervention operation on the path according to the risk state.

Description

Technical Field

[0001] This application relates to the field of deep foundation pit monitoring technology, and in particular to a method, system, medium and product for multi-parameter intelligent monitoring of deep foundation pits based on the Internet of Things. Background Technology

[0002] Deep foundation pit engineering is a crucial component of urban underground space development, and safety monitoring during its construction is of paramount importance for preventing collapse accidents. During deep foundation pit excavation, the interplay of factors such as soil stress redistribution, groundwater seepage, and deformation of the support structure leads to complex dynamic changes in parameters around the pit, including displacement, stress, and water pressure. This multi-parameter coupling often results in the hidden accumulation of potential hazards even when all monitoring parameters are within acceptable limits. By the time a single point of data shows significant anomalies, the safety margin of the foundation pit has been drastically reduced, potentially leading to collapses or settlement of surrounding buildings.

[0003] In related technologies, to address the aforementioned technical problems, an IoT-based intelligent monitoring method for deep foundation pits has been proposed. This method involves deploying a network of various types of sensors around the foundation pit to automatically collect real-time data on multiple physical parameters such as displacement, tilt, axial force, and pore water pressure. A time series prediction model is then used to analyze the trends of the data at each monitoring point. Specifically, a long short-term memory network is used to model and learn from historical data at each monitoring point, predicting the development trends of each parameter over a future period. When the predicted value exceeds a set threshold, an early warning is triggered, thereby improving the frequency of data collection and the timeliness of early warnings to a certain extent.

[0004] However, the aforementioned IoT-based intelligent monitoring method for deep foundation pits only makes independent predictions of the numerical change trends of each monitoring point. It is difficult to effectively identify the cumulative effect of systemic risks caused by small changes in multiple monitoring parameters within the safety threshold range. Therefore, even if the predicted values ​​of each measuring point are within the safe range, the deep foundation pit may still experience sudden emergencies due to the accumulation of implicit correlations between multiple parameters, resulting in poor reliability of multi-parameter collaborative early warning for deep foundation pits in related technologies. Summary of the Invention

[0005] This application provides a method, system, medium, and product for intelligent monitoring of multiple parameters in deep foundation pits based on the Internet of Things, which can improve the reliability of collaborative early warning of multiple parameters in deep foundation pits.

[0006] Firstly, this application provides an IoT-based intelligent monitoring method for multiple parameters of deep foundation pits, applied to the aforementioned intelligent monitoring system for multiple parameters of deep foundation pits. The method includes: upon acquiring multiple types of monitoring data of the deep foundation pit, determining the cross-physical quantity transmission relationships between various monitoring parameters in the multiple types of monitoring data, and constructing a heterogeneous correlation graph structure based on these cross-physical quantity transmission relationships. The heterogeneous correlation graph structure includes monitoring nodes and influencing factor nodes. Each monitoring node corresponds one-to-one with a monitoring parameter, and each influencing factor node corresponds one-to-one with an external influencing event of the deep foundation pit. The correlation edges in the heterogeneous correlation graph structure represent the physical transmission paths between monitoring nodes and between monitoring nodes and influencing factor nodes. Spatiotemporal transmission features are extracted from the multiple types of monitoring data on the heterogeneous correlation graph structure to obtain the temporal variation features of the monitoring nodes. The transmission influence characteristics of associated edges are analyzed, and dynamic weight perturbation operations are performed on associated edges based on temporal change characteristics and transmission influence characteristics to obtain the weight stability index of associated edges. Associated edges in heterogeneous association graph structures with weight stability indices greater than a preset stability threshold are marked as effective transmission edges, and a set of stable transmission paths is determined based on the connectivity of effective transmission edges in the heterogeneous association graph structure. When there is a target transmission path in the set of stable transmission paths that meets the collaborative risk triggering condition, the monitoring nodes in the target transmission path are determined to be in a multi-parameter collaborative risk state, and risk response intervention operations are performed on the target transmission path based on the multi-parameter collaborative risk state. The collaborative risk triggering condition is that the total number of monitoring nodes in the target transmission path is greater than a preset node number threshold and the target transmission path starts with an influencing factor node.

[0007] By adopting the above technical solution, a heterogeneous correlation graph structure integrating physical constraints is constructed, unifying the cross-physical quantity transmission relationships between multiple types of monitoring data and the effects of external influencing events on monitoring parameters. The stability of the correlation edges is verified through dynamic weight perturbation operations, and reliable and effective transmission edges are selected. Stable transmission paths are identified based on the connectivity of effective transmission edges, achieving a paradigm shift from single-point threshold judgment to path collaborative judgment in early warning. This solves the technical problem of poor reliability in multi-parameter collaborative early warning for deep foundation pits in related technologies, achieving the technical effect of improving the reliability of multi-parameter collaborative early warning for deep foundation pits.

[0008] Optionally, when multiple types of monitoring data for the deep foundation pit are obtained, the cross-physical quantity transmission relationships between the monitoring parameters in the multiple types of monitoring data are determined, and a heterogeneous correlation diagram structure is constructed based on the cross-physical quantity transmission relationships. Specifically, this includes: obtaining multiple types of monitoring data, including displacement monitoring data, stress monitoring data, water pressure monitoring data, and tilt monitoring data; obtaining the distribution of the support layers, the connection relationship of the retaining wall, the influence range of the dewatering wells, and the monitoring location information of each monitoring parameter in the deep foundation pit; and performing physical quantity type matching analysis on the multiple types of monitoring data to determine the cross-physical quantity transmission relationships between different types of monitoring parameters. Based on the distribution of supporting layers, the connection relationship of retaining walls, and the influence range of dewatering wells, the structural force transmission path between each monitoring parameter is determined, and the monitoring spatial distance between each monitoring parameter is determined based on the monitoring location information; the monitoring spatial distance is mapped to a preset distance influence function to obtain a distance weight coefficient that characterizes the degree of spatial proximity; a physical transmission constraint matrix is ​​generated based on the cross-physical quantity transmission relationship, the structural force transmission path, and the distance weight coefficient, and the element values ​​in the physical transmission constraint matrix characterize the transmission intensity coefficient between different types of monitoring parameters; a heterogeneous correlation graph structure is constructed based on the physical transmission constraint matrix, monitoring nodes, and influence factor nodes.

[0009] By adopting the above technical solutions, it is possible to obtain multiple types of monitoring data such as displacement, stress, water pressure, and tilt of deep foundation pits. Combined with the distribution of support layers, the connection relationship of retaining walls, and the influence range of dewatering wells, the structural force transmission path between various monitoring parameters can be determined. Through physical quantity type matching analysis, the cross-physical quantity transmission relationship between different types of monitoring parameters can be determined. Based on the monitoring location information, the distance weight coefficient is calculated to generate a physical transmission constraint matrix, providing a scientific physical basis for the construction of heterogeneous correlation graph structure, ensuring that the graph structure can accurately reflect the real physical relationship between various monitoring parameters of deep foundation pits.

[0010] Optionally, a heterogeneous correlation graph structure is constructed based on the physical transmission constraint matrix, monitoring nodes, and impact factor nodes. Specifically, this includes: acquiring construction impact events and environmental impact events for each monitoring parameter; creating monitoring nodes for each monitoring parameter, construction factor nodes for construction impact events, and environmental factor nodes for environmental impact events; performing an effectiveness comparison analysis between the transmission intensity coefficient and a preset transmission threshold to identify two monitoring parameters with a transmission intensity coefficient greater than the preset transmission threshold as monitoring parameter pairs with an effective transmission relationship; determining the spatial range of construction impact based on the event type and construction location of the construction impact event, and determining the spatial range of environmental impact based on the event type and area of ​​effect of the environmental impact event; and performing a first spatial location matching between the spatial range of construction impact and the monitoring location information to obtain a first... The matching results are used to determine the monitoring nodes whose monitoring location information is within the construction impact space range as construction-affected nodes. The degree of construction transmission impact is determined based on the monitoring location information of the construction-affected nodes and the corresponding monitoring location information of the construction factor nodes. The environmental impact space range is matched with the monitoring location information to obtain the second matching result. The monitoring nodes whose monitoring location information is within the environmental impact space range are determined based on the second matching result. The degree of environmental transmission impact is determined based on the monitoring location information of the environmental-affected nodes and the corresponding monitoring location information of the environmental factor nodes. A heterogeneous association graph structure is constructed based on the monitoring parameter pairs, construction-affected nodes, degree of construction transmission impact, environmental-affected nodes, and environmental-affected nodes.

[0011] By adopting the above technical solution, construction factor nodes and environmental factor nodes are created based on construction impact events and environmental impact events, respectively. Through effectiveness comparison analysis, monitoring parameter pairs with effective transmission relationships are determined. Based on the spatial range of construction impact and environmental impact, spatial location matching is performed to determine the affected nodes. The correlation between impact factor nodes and monitoring nodes is established according to the degree of construction transmission impact and environmental transmission impact. This fully expresses the multi-source heterogeneous correlation relationship in the deep foundation pit monitoring system, providing a structured data foundation for subsequent collaborative risk analysis.

[0012] Optionally, spatiotemporal transmission features are extracted from multi-type monitoring data on a heterogeneous correlation graph structure to obtain the temporal change features of monitoring nodes and the transmission influence features of correlation edges. Specifically, this includes: acquiring multiple time-series data segments of monitoring nodes within a preset historical time window; inputting these multiple time-series data segments into a temporal feature encoding network to obtain the temporal change features of monitoring nodes; determining the first change feature of the source node connected to the target correlation edge and the second change feature of the target node connected to the target correlation edge from the temporal change features, and determining the cross-correlation function between the source node and the target node based on the first and second change features; determining the temporal correlation coefficient between the source node and the target node based on the cross-correlation function, and... The conduction delay parameter is determined based on the peak time position of the cross-correlation function and the temporal correlation coefficient. The conduction activity index between the source node and the target node is determined based on the temporal correlation coefficient and the conduction delay parameter. The conduction activity index is then coupled with the initial edge weight of the target associated edge to obtain the conduction response intensity value of the associated edge. The first physical quantity type of the monitoring parameter corresponding to the source node and the second physical quantity type of the monitoring parameter corresponding to the target node are obtained. A cross-physical quantity conduction attenuation factor combining the physical quantity types corresponding to the first and second physical quantity types is queried from a preset physical quantity type mapping table. The conduction response intensity value is modulated with the cross-physical quantity conduction attenuation factor to obtain the conduction influence characteristics.

[0013] By adopting the above technical solution, based on multiple time-series data segments within a preset historical time window, the temporal change characteristics of monitoring nodes are extracted using a temporal feature coding network; the temporal correlation coefficient and conduction delay parameter between the source node and the target node are determined through cross-correlation function analysis, thereby determining the conduction activity index and conduction response intensity value; and a cross-physical quantity conduction attenuation factor is introduced to modulate the conduction effect between different types of physical quantities, obtaining the conduction influence characteristics of the associated edge, which can accurately characterize the change pattern of monitoring parameters in the time dimension and the conduction response characteristics between parameters.

[0014] Optionally, a dynamic weight perturbation operation is performed on the associated edges based on temporal variation characteristics and propagation influence characteristics to obtain the weight stability index of the associated edges. Specifically, this includes: determining the first activity index of the source node based on temporal variation characteristics, and determining the second activity index of the target node based on temporal variation characteristics; determining the edge activation factor of the target associated edge based on the first and second activity indices, and generating the perturbation amplitude coefficient of the target associated edge based on the edge activation factor and propagation influence characteristics; generating multiple sets of random perturbation vectors according to a preset perturbation sampling strategy, and scaling the multiple sets of random perturbation vectors with the perturbation amplitude coefficient to obtain multiple sets of target perturbation amounts; and superimposing the multiple sets of target perturbation amounts with the target initial edge weights to obtain multiple sets of... The algorithm perturbs edge weights and updates multiple sets of perturbation edge weights to the heterogeneous graph structure to determine the change in graph topology connectivity when edge weight perturbations occur. It then performs a sensitivity comparison analysis between the change in graph topology connectivity and a preset connectivity sensitivity threshold to obtain the comparison results. When the sensitivity comparison results indicate that the change in graph topology connectivity is greater than the connectivity sensitivity threshold, a sensitive connectivity label is assigned to the corresponding perturbation edge weight. When the sensitivity comparison results indicate that the change in graph topology connectivity is less than or equal to the connectivity sensitivity threshold, a stable connectivity label is assigned to the corresponding perturbation edge weight. Finally, the weight stability index of the target associated edge is determined based on the number of perturbation edge weights with stable connectivity labels among the multiple sets of perturbation edge weights.

[0015] By adopting the above technical solution, the activity indicators of source and target nodes are determined based on temporal variation characteristics, thereby determining the edge activation factor and perturbation amplitude coefficient. Multiple sets of random perturbation vectors are generated according to a preset perturbation sampling strategy, and dynamic weight perturbation operations are performed on the target associated edges, and the change in graph topological connectivity is analyzed. Sensitive connectivity labels or stable connectivity labels are assigned to the perturbation edge weights through sensitivity comparison analysis. The weight stability index is determined based on the number of perturbation edge weights with stable connectivity labels. This can effectively evaluate the contribution of each associated edge to maintaining the stability of the graph structure, thereby distinguishing the real physical transmission relationship from noise and pseudo-correlation in the data, and improving the reliability of subsequent collaborative risk assessment.

[0016] Optionally, edges in the heterogeneous correlation graph structure with weight stability indices greater than a preset stability threshold are marked as valid transmission edges. A set of stable transmission paths is determined based on the connectivity of these valid transmission edges in the heterogeneous correlation graph structure. Specifically, this includes: performing a stability comparison analysis between the weight stability indices of the edges and the preset stability threshold; marking edges with weight stability indices greater than the preset stability threshold as valid transmission edges and edges with weight stability indices less than or equal to the preset stability threshold as unstable edges; removing all unstable edges from the heterogeneous correlation graph structure to obtain a valid transmission subgraph; performing a connected component detection operation in the valid transmission subgraph to identify multiple independent connected sub-regions; using each influencing factor node in the node set of each connected sub-region as the path start node and each monitoring node in the node set of each connected sub-region as the path end node, performing a path search operation to obtain candidate transmission paths; and determining stable transmission paths from the candidate transmission paths based on the weight stability indices of the valid transmission edges in the candidate transmission paths, and generating a set of stable transmission paths.

[0017] By adopting the above technical solution, based on stability comparison analysis, associated edges with weight stability indices greater than a preset stability threshold are marked as effective transmission edges, while associated edges with weight stability indices less than or equal to the preset stability threshold are marked as unstable edges and removed from the heterogeneous association graph structure, thus obtaining an effective transmission subgraph. In the effective transmission subgraph, multiple connected sub-regions are identified through connected component detection, and a path search operation is performed with the influencing factor node as the path start node and the monitoring node as the path end node to obtain candidate transmission paths, thereby determining a set of stable transmission paths. This can extract highly reliable risk transmission channels from complex heterogeneous association graph structures, providing a reliable path basis for collaborative risk assessment.

[0018] Optionally, when a target transmission path that meets the collaborative risk triggering condition exists in the set of stable transmission paths, the monitoring nodes in the target transmission path are determined to be in a multi-parameter collaborative risk state. Risk response intervention operations are then performed on the target transmission path based on the multi-parameter collaborative risk state. Specifically, this includes: obtaining the total number of monitoring nodes and the total number of influencing factor nodes included in each stable transmission path in the set of stable transmission paths; identifying stable transmission paths with a total number of monitoring nodes greater than a preset node number threshold and a total number of influencing factor nodes greater than or equal to a preset influencing factor number threshold as target transmission paths that meet the collaborative risk triggering condition; obtaining real-time monitoring data for each monitoring node in the target transmission path, and determining the degree of data deviation for each monitoring node in the target transmission path based on the real-time monitoring data. Based on the degree of data deviation, determine whether each monitoring node in the target transmission path is in a multi-parameter collaborative risk state; if it is determined that each monitoring node in the target transmission path is in a multi-parameter collaborative risk state, determine the path collaborative risk index based on the degree of data deviation and the location information of each monitoring node in the target transmission path; if it is determined that the target transmission path is at the early warning observation risk level based on the path collaborative risk index, execute the early warning observation task for the target transmission path; if it is determined that the target transmission path is at the encrypted monitoring risk level based on the path collaborative risk index, execute the encrypted collaborative monitoring task for the target transmission path; if it is determined that the target transmission path is at the emergency intervention risk level based on the path collaborative risk index, execute the risk blocking task for the target transmission path.

[0019] By adopting the above technical solution, the target transmission path that meets the conditions for triggering collaborative risks is determined based on the total number of monitoring nodes and the total number of influencing factor nodes. The data deviation of each monitoring node is determined based on real-time monitoring data, and it is judged whether it is in a multi-parameter collaborative risk state. The path collaborative risk index is determined based on the data deviation and the location information of the monitoring nodes. Based on the path collaborative risk index, three risk response levels are divided: early warning observation risk level, intensified monitoring risk level, and emergency intervention risk level. Early warning observation tasks, intensified collaborative monitoring tasks, and risk blocking tasks are executed for different risk response levels, respectively, to realize intelligent hierarchical response of deep foundation pit safety monitoring, and maximize construction safety while ensuring monitoring efficiency.

[0020] Secondly, embodiments of this application provide a multi-parameter intelligent monitoring system for deep foundation pits. The multi-parameter intelligent monitoring system for deep foundation pits includes: one or more processors and a memory; the memory is coupled to one or more processors, and the memory is used to store computer program code, the computer program code including computer instructions, and the one or more processors call the computer instructions to cause the multi-parameter intelligent monitoring system for deep foundation pits to perform the method described in the first aspect and any possible implementation of the first aspect.

[0021] Thirdly, embodiments of this application provide a computer program product containing instructions that, when the computer program product is run on a deep foundation pit multi-parameter intelligent monitoring system, cause the deep foundation pit multi-parameter intelligent monitoring system to execute the method described in the first aspect and any possible implementation thereof.

[0022] Fourthly, embodiments of this application provide a computer-readable storage medium including instructions that, when executed on a deep foundation pit multi-parameter intelligent monitoring system, cause the deep foundation pit multi-parameter intelligent monitoring system to perform the method described in the first aspect and any possible implementation thereof. Attached Figure Description

[0023] Figure 1 This is a flowchart illustrating a multi-parameter intelligent monitoring method for deep foundation pits based on the Internet of Things in an embodiment of this application. Figure 2 This is a schematic diagram of the physical device structure of a multi-parameter intelligent monitoring system for deep foundation pits in this application embodiment. Detailed Implementation

[0024] The terminology used in the following embodiments of this application is for the purpose of describing particular embodiments only and is not intended to be limiting of this application. As used in the specification and appended claims of this application, the singular expressions “a,” “an,” “the,” “the,” “the,” and “this” are intended to include the plural expressions as well, unless the context clearly indicates otherwise. It should also be understood that the term “and / or” as used in this application refers to any or all possible combinations including one or more of the listed items.

[0025] Hereinafter, the terms "first" and "second" are used for descriptive purposes only and should not be construed as implying or suggesting relative importance or implicitly indicating the number of indicated technical features. Thus, a feature defined as "first" or "second" may explicitly or implicitly include one or more of that feature, and in the description of the embodiments of this application, unless otherwise stated, "multiple" means two or more.

[0026] This application provides a multi-parameter intelligent monitoring method for deep foundation pits based on the Internet of Things (IoT). (See attached document.) Figure 1 , Figure 1 This is a flowchart illustrating a multi-parameter intelligent monitoring method for deep foundation pits based on the Internet of Things, as described in this application, including the following steps: Step S101: In the case of obtaining multi-type monitoring data of deep foundation pit, determine the cross-physical quantity transmission relationship between each monitoring parameter in the multi-type monitoring data, and construct a heterogeneous correlation graph structure based on the cross-physical quantity transmission relationship. The heterogeneous correlation graph structure includes monitoring nodes and influencing factor nodes. The monitoring nodes correspond one-to-one with each monitoring parameter, and the influencing factor nodes correspond one-to-one with the external influencing events of deep foundation pit. The correlation edges included in the heterogeneous correlation graph structure represent the physical transmission paths between monitoring nodes and between monitoring nodes and influencing factor nodes. Step S102: Extract the spatiotemporal transmission features of multiple types of monitoring data on the heterogeneous association graph structure to obtain the temporal change features of monitoring nodes and the transmission influence features of association edges. Then, perform dynamic weight perturbation operation on the association edges according to the temporal change features and transmission influence features to obtain the weight stability index of the association edges. Step S103: Mark the edges in the heterogeneous association graph structure whose weight stability index is greater than the preset stability threshold as effective transmission edges, and determine the set of stable transmission paths based on the connectivity of the effective transmission edges in the heterogeneous association graph structure. Step S104: When there is a target transmission path in the set of stable transmission paths that meets the conditions for triggering collaborative risks, determine that the monitoring nodes in the target transmission path are in a multi-parameter collaborative risk state, and perform risk response intervention operations on the target transmission path according to the multi-parameter collaborative risk state. The conditions for triggering collaborative risks are that the total number of monitoring nodes in the target transmission path is greater than the preset node number threshold and the target transmission path starts with the influencing factor node.

[0027] In the above embodiments, the deep foundation pit multi-parameter intelligent monitoring system acquires various types of monitoring data through an Internet of Things (IoT) sensor network. Specifically, the various types of monitoring data may include, but are not limited to: horizontal displacement data of the retaining structure, axial force data of the supports, groundwater level data, surrounding surface settlement data, and tilt data of the retaining wall. Cross-physical quantity transmission relationship refers to the causal relationship between different physical quantities based on the principles of engineering mechanics. For example, when dewatering in the foundation pit causes a drop in the groundwater level (water pressure change), it will cause an increase in the effective stress of the soil, which in turn causes the retaining structure to undergo horizontal displacement (displacement change) towards the inside of the foundation pit. This correlation between water pressure and displacement is a cross-physical quantity transmission relationship. The heterogeneous correlation graph structure G can be represented as G=(V, E, W), where V is the set of nodes, including the set of monitoring nodes Vm and the set of influencing factor nodes Vf; E is the set of edges, representing the physical transmission path between nodes; and W is the set of edge weights, representing the transmission intensity. The preset stability threshold can be set to a value between 0.7 and 0.9. The threshold value can be determined based on the importance level of the project: for Level 1 foundation pit projects, a value of 0.85 is recommended; for Level 2 foundation pit projects, a value of 0.75 is recommended. The preset node number threshold can be set to 3, indicating that when the number of monitoring nodes in the stable transmission path exceeds 3, a multi-parameter collaborative risk is considered to exist. Through the above steps, this embodiment can identify systemic risks from the collaborative change patterns of multiple monitoring parameters, rather than relying solely on threshold judgments based on single-point data, thereby improving the reliability of multi-parameter collaborative early warning for deep foundation pits. When an external impact event (such as heavy rain or nearby construction) triggers risk transmission, even if the data at each monitoring point does not exceed the single-point alarm threshold, the deep foundation pit multi-parameter intelligent monitoring system can still issue a collaborative risk warning in advance by identifying the activation state of the stable transmission path.

[0028] Through the above steps, a heterogeneous correlation graph structure integrating physical constraints is constructed, unifying the cross-physical quantity transmission relationships between various types of monitoring data and the effects of external influencing events on monitoring parameters. The stability of the correlation edges is verified through dynamic weight perturbation operations, and reliable and effective transmission edges are selected. Stable transmission paths are identified based on the connectivity of effective transmission edges, achieving a paradigm shift from single-point threshold judgment to path-based collaborative judgment in early warning. This solves the technical problem of poor reliability in multi-parameter collaborative early warning for deep foundation pits in related technologies, achieving the technical effect of improving the reliability of multi-parameter collaborative early warning for deep foundation pits.

[0029] The entity performing the above steps may be a system, a device, a controller or processor in a device or system, a standalone controller or processor, or other processing devices or processing units with similar processing functions, but is not limited to these.

[0030] In an optional embodiment, when multiple types of monitoring data of the deep foundation pit are obtained, the cross-physical quantity transmission relationship between each monitoring parameter in the multiple types of monitoring data is determined, and a heterogeneous correlation graph structure is constructed based on the cross-physical quantity transmission relationship. Specifically, this includes: obtaining displacement monitoring data, stress monitoring data, water pressure monitoring data, and tilt monitoring data of the deep foundation pit, and using the displacement monitoring data, stress monitoring data, water pressure monitoring data, and tilt monitoring data as multiple types of monitoring data; obtaining the support structure design information of the deep foundation pit and the monitoring location information of each monitoring parameter, the support structure design information including the distribution of support layers, the connection relationship of the retaining wall, and the influence range of the dewatering well; and performing physical quantity type matching analysis on the displacement monitoring data, stress monitoring data, water pressure monitoring data, and tilt monitoring data to determine the cross-physical quantity transmission relationship between different types of monitoring parameters. The transmission relationships between physical quantities include the seepage deformation coupling relationship between water pressure monitoring data and displacement monitoring data, and the mechanical deformation transmission relationship between stress monitoring data and tilt monitoring data. Based on the distribution of support layers, the connection relationship of the retaining wall, and the influence range of dewatering wells, the structural force transmission path between each monitoring parameter is determined. The monitoring spatial distance between each monitoring parameter is determined based on the monitoring location information, and the monitoring spatial distance is mapped to a preset distance influence function to obtain a distance weight coefficient characterizing the degree of spatial proximity. A physical transmission constraint matrix is ​​generated based on the transmission relationships between physical quantities, the structural force transmission path, and the distance weight coefficient. The element values ​​in the physical transmission constraint matrix characterize the transmission strength coefficient between different types of monitoring parameters. A heterogeneous correlation graph structure is constructed based on the physical transmission constraint matrix, monitoring nodes, and influence factor nodes.

[0031] In the above embodiments, displacement monitoring data is acquired using an inclinometer and displacement gauge installed on the top of the retaining structure. The data is collected every 2 hours, and the data format is (measuring point ID, timestamp, X-direction displacement value, Y-direction displacement value, Z-direction displacement value), with the unit being millimeters. Stress monitoring data is acquired using an axial force gauge installed on the steel support, collected every 1 hour, with the data format being (measuring point ID, timestamp, axial force value), with the unit being kilonewtons. Water pressure monitoring data is acquired using a water level gauge buried in an underground water level observation well, collected every 30 minutes, with the data format being (measuring point ID, timestamp, water level elevation), with the unit being meters. Inclination monitoring data is acquired using an inclinometer installed on the retaining wall, collected every 2 hours, with the data format being (measuring point ID, timestamp, inclination angle), with the unit being degrees. The support structure design information can be extracted from the engineering BIM model or CAD design drawings, specifically including: support layer distribution information, recording the elevation position of each support, such as the first support being located at -2.0m, the second support at -6.5m, and the third support at -11.0m; the connection relationship of the retaining wall, recording the segment information of the retaining wall and the connection method of adjacent wall segments; the influence range of the dewatering wells: based on the design output of the dewatering wells and the soil permeability coefficient, the effective influence radius of each dewatering well is calculated, which is usually 15-30 meters.

[0032] In the above embodiments, the physical mechanism of the cross-physical quantity transmission relationship is as follows: Seepage-deformation coupling relationship: Based on Terzaghi's effective stress principle, the total soil stress is equal to the sum of the effective stress and the pore water pressure (σ=σ'+u). When the pore water pressure u decreases due to dewatering of the foundation pit, the effective stress σ' increases, which in turn causes soil compression and displacement of the retaining structure. This coupling relationship can be represented by the transmission coefficient α_hw, and its calculation formula is α_hw=k_s×m_v / (1+e_0), where k_s is the soil permeability coefficient, m_v is the soil compression coefficient, and e_0 is the initial void ratio. Mechanical deformation transmission relationship: Based on the principle of structural mechanical equilibrium, when a certain part of the retaining structure deforms, the load will be transmitted to the adjacent parts through the support system. The strength of this transmission relationship is related to the support stiffness and structural continuity. The preset distance influence function adopts the Gaussian decay function form w(d)=exp(-d² / 2σ²), where d is the Euclidean distance between the two monitoring points (unit: meters), and σ is the distance decay parameter. The value of σ can be determined according to the size of the foundation pit: for small foundation pits with a side length of less than 50 meters, it is recommended that σ be 8-12 meters; for medium-sized foundation pits with a side length of 50-100 meters, it is recommended that σ be 12-18 meters; for large foundation pits with a side length of more than 100 meters, it is recommended that σ be 18-25 meters.

[0033] In the above embodiments, the preset distance influence function can also adopt the linear decay form w(d)=max(0,1-d / d_max), where d_max is the maximum influence distance, and the spatial proximity is considered to be 0 if the distance exceeds this. The physical transmission constraint matrix M is constructed as follows: matrix M is an n×n square matrix, where n is the total number of monitoring parameters. The matrix element M[i][j] represents the transmission strength coefficient between the i-th monitoring parameter and the j-th monitoring parameter, and its calculation formula is M[i][j]=β_type×γ_struct×w(d_ij), where: β_type is the basic coefficient of cross-physical quantity transmission, which is determined by the combination of physical quantity types of monitoring parameters i and j; γ_struct is the structural force transmission coefficient, which takes a value of 1.0 when i and j are located in the same support layer or the same retaining wall segment, and otherwise takes a value of 0.3-0.8 according to the degree of structural connection; w(d_ij) is the distance weight coefficient, which is calculated by the preset distance influence function.

[0034] In an optional embodiment, a heterogeneous correlation graph structure is constructed based on the physical transmission constraint matrix, monitoring nodes, and impact factor nodes. Specifically, this includes: acquiring construction impact events and environmental impact events for each monitoring parameter, where external impact events include both construction impact events and environmental impact events; creating monitoring nodes for each monitoring parameter, creating construction factor nodes for construction impact events, and creating environmental factor nodes for environmental impact events, where impact factor nodes include both construction factor nodes and environmental factor nodes; performing an effectiveness comparison analysis between the transmission intensity coefficient and a preset transmission threshold to obtain effectiveness comparison results, and determining two monitoring parameters with transmission intensity coefficients greater than the preset transmission threshold as monitoring parameter pairs with effective transmission relationships based on the effectiveness comparison results; establishing a first correlation edge between the monitoring nodes corresponding to the monitoring parameter pairs, and assigning a first initial edge weight to the first correlation edge based on the transmission intensity coefficient; determining the spatial range of construction impact based on the event type and construction location of the construction impact event, and determining the spatial range of environmental impact based on the event type and area of ​​effect of the environmental impact event; performing a first spatial location matching between the spatial range of construction impact and the monitoring location information to obtain a first matching result, and then assigning the monitoring location information to the monitoring parameter pairs based on the first matching result. Monitoring nodes located within the construction impact space are identified as construction-affected nodes. A second association edge is established between the construction factor nodes corresponding to the construction impact events that generate the construction impact space and the construction-affected nodes. The degree of construction transmission impact is determined based on the monitoring location information of the construction-affected nodes and the corresponding monitoring factor nodes. A second initial edge weight is assigned to the second association edge based on the degree of construction transmission impact. The environmental impact space is matched with the monitoring location information to obtain a second matching result. Monitoring nodes whose monitoring location information is located within the environmental impact space are identified as environmental-affected nodes. A third association edge is established between the environmental factor nodes corresponding to the environmental impact events that generate the environmental impact space and the environmental-affected nodes. The degree of environmental transmission impact is determined based on the monitoring location information of the environmental-affected nodes and the corresponding monitoring factor nodes. A third initial edge weight is assigned to the third association edge based on the degree of environmental transmission impact. A heterogeneous association graph structure is constructed based on the monitoring nodes, impact factor nodes, first association edge, second association edge, third association edge, first initial edge weight, second initial edge weight, and third initial edge weight.

[0035] In the above embodiments, external impact events are divided into two categories: construction impact events and environmental impact events. Construction impact events include, but are not limited to: foundation pit excavation events (recording excavation location, depth, and time); support erection / removal events (recording support number and erection / removal time); adjacent pile foundation construction events (recording pile location coordinates, construction time, and pile diameter); grouting reinforcement events (recording grouting location, grouting volume, and construction time); and heavy vehicle passage events (recording passage route, vehicle load, and passage time). Environmental impact events include, but are not limited to: rainfall events (recording rainfall start time, duration, and rainfall level (light rain / moderate rain / heavy rain / torrential rain)); abnormal groundwater level changes (recording change amplitude and rate of change); surrounding building settlement events (recording settlement location and amount); and sudden temperature change events (recording temperature change amplitude and time period). A preset conduction threshold is used to filter monitoring parameter pairs with effective conduction relationships. In this embodiment, the preset conduction threshold can be set between 0.3 and 0.5. When the element value in the physical transmission constraint matrix is ​​greater than the preset transmission threshold, a valid physical transmission relationship is considered to exist between the two corresponding monitoring parameters. The spatial range of construction impact is determined according to the construction type: Earthwork excavation: the impact area is the region formed by extending the excavation boundary by twice the excavation depth; Pile foundation construction: the impact area is a circular area with a radius of three times the pile diameter at the pile center; Heavy vehicles: the impact area is a strip-shaped area 10 meters on each side of the traffic route. The spatial range of environmental impact is determined according to the event type: Rainfall events: the impact area is the entire foundation pit area; Groundwater level anomalies: the impact area is a circular area with a radius of 30 meters at the anomaly point.

[0036] In the above embodiments, the calculation method for the degree of construction transmission impact is as follows: P_construction=E_level×(1-d_c / R_c), where E_level is the construction event level coefficient, with a value range of 0.5-1.0 (0.5 for general construction, 0.7 for important construction, and 1.0 for critical construction); d_c is the distance from the monitoring node to the center of the construction impact event; R_c is the radius of the construction impact space range; when d_c≥R_c, P_construction=0. The calculation method for the degree of environmental transmission impact is as follows: P_environment=S_level×exp(-d_e / λ), where S_level is the environmental event severity coefficient (0.3 for light rain, 0.5 for moderate rain, 0.7 for heavy rain, and 1.0 for torrential rain); d_e is the distance from the monitoring node to the center of the environmental impact event; λ is the environmental impact attenuation constant, generally with a value of 10-20 meters. The establishment rules and weight allocation for the three types of associated edges are as follows: First associated edge (monitoring node - monitoring node): The establishment condition is that the corresponding element value in the physical transmission constraint matrix > the preset transmission threshold; the first initial edge weight = transmission intensity coefficient (i.e., matrix element value). Second associated edge (construction factor node - monitoring node): The establishment condition is that the monitoring node is located within the construction influence space; the second initial edge weight = construction transmission influence degree P_construction. Third associated edge (environmental factor node - monitoring node): The establishment condition is that the monitoring node is located within the environmental influence space; the third initial edge weight = environmental transmission influence degree P_environment. The heterogeneous association graph structure constructed in the above manner can fully express the physical transmission relationship between monitoring parameters and the effect of external influence events on monitoring parameters.

[0037] In an optional embodiment, spatiotemporal transmission features are extracted from multiple types of monitoring data on a heterogeneous correlation graph structure to obtain the temporal change features of monitoring nodes and the transmission influence features of correlation edges. Specifically, this includes: acquiring historical monitoring sequence data of monitoring nodes within a preset historical time window, and performing sliding window segmentation on the historical monitoring sequence data to obtain multiple time-series data segments; inputting the multiple time-series data segments into a time-series feature encoding network for time-series feature encoding processing to obtain node-level time-series embedding vectors of monitoring nodes, which represent the monitoring value change pattern of monitoring nodes within the preset historical time window; determining the rate of change and acceleration of change features of monitoring nodes based on the node-level time-series embedding vectors, and concatenating the rate of change and acceleration of change features to obtain the temporal change features of monitoring nodes; determining the first change feature of the source node connected to the target correlation edge and the second change feature of the target node connected to the target correlation edge from the temporal change features, and determining the source node and target node based on the first change feature and the second change feature. The cross-correlation function is used to determine the temporal correlation coefficient between the source node and the target node. The peak time position of the cross-correlation function and the temporal correlation coefficient are used to determine the propagation delay parameter, which characterizes the time lag in the propagation of the source node's change signal to the target node. The propagation activity index between the source node and the target node is determined based on the temporal correlation coefficient and the propagation delay parameter. This index is then coupled with the initial edge weight of the target associated edge to obtain the propagation response intensity value of the associated edge. The first physical quantity type of the monitoring parameter corresponding to the source node and the second physical quantity type of the monitoring parameter corresponding to the target node are obtained. A cross-physical quantity propagation attenuation factor corresponding to the first and second physical quantity types is queried from a preset physical quantity type mapping table. The preset physical quantity type mapping table pre-stores different combinations of physical quantity types and the cross-physical quantity propagation attenuation factors corresponding to different combinations of physical quantity types. The propagation response intensity value is modulated with the cross-physical quantity propagation attenuation factor to obtain the propagation influence characteristics.

[0038] In the above embodiments, the length of the preset historical time window can be determined according to the foundation pit monitoring cycle. In this embodiment, the preset historical time window can be set to 72 hours (3 days) to capture the short-term changing trend of the monitoring data. The parameter configuration for the sliding window segmentation process is as follows: window size: 24 sampling points (equivalent to 48 hours for data collected every 2 hours); sliding step size: 6 sampling points (equivalent to 12 hours); number of time series data segments obtained after segmentation = (72-48) / 12+1 = 3 segments. The temporal feature encoding network adopts an LSTM (Long Short-Term Memory) structure, with the following specific configuration: Input layer: receives temporal data with dimensions (batch_size, sequence_length, feature_dim), where sequence_length=24 and feature_dim=1; First LSTM layer: has 128 hidden units and returns the complete sequence; Dropout layer: has a dropout rate of 0.2 to prevent overfitting; Second LSTM layer: has 64 hidden units and returns only the hidden state at the last time step; Fully connected layer: maps the 64-dimensional hidden state to a 32-dimensional node-level temporal embedding vector.

[0039] In the above embodiments, the temporal feature encoding network can also adopt a Transformer encoder structure or a temporal convolutional network (TCN) structure (the temporal feature encoding network can also adopt a GRU (gated recurrent unit) network structure). The node-level temporal embedding vector has a dimension of 32. This vector can characterize the monitoring value change pattern of the monitoring node within the historical time window, including features such as trend, periodicity, and volatility. The calculation formulas for the rate of change feature and the acceleration of change feature are as follows: rate of change feature v = (x_t - x_{t - Δt}) / Δt; acceleration of change feature a = (v_t - v_{t - Δt}) / Δt, where x_t is the monitoring value at the current moment, and Δt is the sampling time interval. The rate of change feature (32-dimensional) and the acceleration of change feature (32-dimensional) are concatenated to obtain a 64-dimensional temporal change feature vector. The cross-correlation function describes the variation of the correlation between two time series with time delay. Its mathematical definition is R_xy(τ) = Σ[x(t)×y(t+τ)] / √[Σx²(t)×Σy²(t+τ)], where τ is the time delay, and R_xy(τ) ranges from -1 to 1. The time series correlation coefficient is defined as the maximum absolute value of the cross-correlation function ρ_xy = max|R_xy(τ)|, ranging from 0 to 1. The propagation delay parameter τ_d is defined as the time delay τ_d = argmax|R_xy(τ)| that maximizes the cross-correlation function. This parameter represents the time required for the changed signal from the source node to propagate to the target node, expressed in hours.

[0040] In the above embodiments, in deep foundation pit engineering, the conduction delay parameter is typically within the range of 0-24 hours. The formula for calculating the conduction activity index A_xy is A_xy=ρ_xy×exp(-τ_d / T_ref), where ρ_xy is the temporal correlation coefficient; τ_d is the conduction delay parameter; and T_ref is the reference time constant, which can be 12 hours. The larger the conduction activity index, the more active the conduction relationship between the two nodes. The formula for calculating the conduction response intensity value S_xy is S_xy=A_xy×W_0, where W_0 is the target initial edge weight of the target associated edge. The preset physical quantity types and cross-physical quantity conduction attenuation factors are shown in Table 1 below: Table 1 Preset Physical Quantity Type Mapping Table The formula for calculating the conduction influence characteristic F_xy is F_xy=S_xy×η_type, where η_type is the cross-physical quantity conduction attenuation factor obtained from the preset physical quantity type mapping table. The value of the cross-physical quantity conduction attenuation factor is determined based on the practical experience of deep foundation pit engineering and can be adjusted according to specific engineering conditions in actual applications.

[0041] In an optional embodiment, a dynamic weight perturbation operation is performed on the associated edges based on temporal variation characteristics and propagation influence characteristics to obtain a weight stability index for the associated edges. Specifically, this includes: determining the monitoring activity index of the monitoring nodes based on temporal variation characteristics, whereby the monitoring activity index characterizes the data fluctuation intensity of the monitoring nodes during the current monitoring period; determining a first activity index of the source node from the monitoring activity index, and determining a second activity index of the target node from the monitoring activity index; determining the edge activation factor of the target associated edge based on the first and second activity indices, and generating a perturbation amplitude coefficient of the target associated edge based on the edge activation factor and propagation influence characteristics; generating multiple sets of random perturbation vectors according to a preset perturbation sampling strategy, and scaling the multiple sets of random perturbation vectors with the perturbation amplitude coefficient to obtain multiple sets of target perturbation quantities; and then... Multiple sets of perturbed edge weights are obtained by superimposing them with the initial edge weights of the target edge, and these perturbed edge weights are then updated in the heterogeneous graph structure to determine the change in graph topology connectivity when edge weight perturbation occurs. The change in graph topology connectivity is then compared with a preset connectivity sensitivity threshold to obtain the sensitivity comparison results. When the sensitivity comparison results indicate that the change in graph topology connectivity is greater than the connectivity sensitivity threshold, a sensitive connectivity label is assigned to the corresponding perturbed edge weight. When the sensitivity comparison results indicate that the change in graph topology connectivity is less than or equal to the connectivity sensitivity threshold, a stable connectivity label is assigned to the corresponding perturbed edge weight. The number of perturbed edge weights with stable connectivity labels in the multiple sets of perturbed edge weights is determined, and the ratio of the number of perturbed edge weights to the total number of perturbed edge weights in the multiple sets is used as the weight stability index of the target associated edge.

[0042] In the above embodiments, the specific implementation of the dynamic weight perturbation operation is as follows: The monitoring activity index is used to characterize the data fluctuation intensity of the monitoring node in the current monitoring period. Its calculation formula is Act_i=std(X_i) / mean(|X_i|), where X_i is the monitoring data sequence of monitoring node i in the current monitoring period; std(X_i) is the standard deviation of the data sequence; and mean(|X_i|) is the mean of the absolute values ​​of the data sequence. The larger the monitoring activity index, the more severe the data fluctuation of the node. The edge activation factor is used to characterize the comprehensive activity level of the nodes at both ends of the associated edge. Its calculation formula is AF_xy=(Act_x+Act_y) / 2×(1+|Act_x-Act_y|), where Act_x is the first activity index of the source node, and Act_y is the second activity index of the target node. The greater the difference in activity between the two ends of the node, the higher the edge activation factor. The perturbation amplitude coefficient controls the intensity of the perturbation applied to the edge weights. Its calculation formula is σ_xy = AF_xy × F_xy × k_perturb, where AF_xy is the edge activation factor; F_xy is the propagation influence characteristic; and k_perturb is the perturbation intensity adjustment coefficient, with a recommended value of 0.1-0.3. The preset perturbation sampling strategy uses a Gaussian distribution with a mean of 0: ε ~ N(0,1). Specifically, for each associated edge, N sets of random perturbation vectors are generated. N can take values ​​from 50 to 100 sets to ensure the reliability of the statistical results.

[0043] In the above embodiment, the calculation formula for multiple sets of target perturbation quantities is δ_k=ε_k×σ_xy, where k=1,2,...,N, and the calculation formula for multiple sets of perturbation edge weights is W_k=W_0+δ_k, where W_0 is the initial target edge weight. The quantification method for graph topology connectivity change is as follows: Step 1, for the original heterogeneous graph structure G, the effective edge is determined by the edge weight threshold method. Let the edge weight threshold be W_th (taken as 0.5 times the average of all edge weights). When the weight of an edge is less than W_th, it is considered that the edge is broken. The number of nodes contained in the largest connected component after the broken edge in the original graph structure is counted and denoted as N_original. Step 2, update the weight W_k of the k-th perturbation edge in the graph structure (only update the weight of the target associated edge), and re-execute the calculation of Step 1 to obtain the number of nodes contained in the largest connected component of the perturbation graph structure, denoted as N_perturbed(k). Step 3: The formula for calculating the change in graph topological connectivity ΔC_k is ΔC_k = |N_original - N_perturbed(k)| / N_original, and the value range of this change is [0,1]. The larger ΔC_k is, the more significant the impact of this perturbation on the overall connectivity of the graph, indicating that the target associated edge is more important for maintaining graph connectivity.

[0044] In the above embodiments, the change in graph topology connectivity can also be calculated using the following alternative methods: Method 2 (based on the number of connected components): ΔC_k = |C_original - C_perturbed(k)| / C_original, where C_original and C_perturbed(k) are the number of connected components in the graph before and after the perturbation, respectively. Method 3 (based on average path length): ΔC_k = |L_original - L_perturbed(k)| / L_original, where L_original and L_perturbed(k) are the average shortest path lengths between all node pairs in the graph before and after the perturbation, respectively. A preset connectivity sensitivity threshold is used to determine whether the change in graph topology connectivity is significant. In this embodiment, the preset connectivity sensitivity threshold can be set between 0.1 and 0.2. The selection criteria for this threshold are as follows: when ΔC_k > 0.2, it is considered that the perturbation has caused a significant change in connectivity, and the edge is relatively sensitive to maintaining the stability of the graph structure; when ΔC_k ≤ 0.1, it is considered that the perturbation has not caused a significant change in connectivity, and the edge has good stability. The judgment logic of sensitivity comparison analysis is as follows: if ΔC_k > preset connectivity sensitivity threshold, then sensitive connectivity labels are assigned to the weights of the k-th perturbation edge; if ΔC_k ≤ preset connectivity sensitivity threshold, then stable connectivity labels are assigned to the weights of the k-th perturbation edge. The calculation formula for the weight stability index SI_xy is SI_xy = N_stable / N, where N_stable is the number of perturbation edge weights with stable connectivity labels; N is the total number of perturbation edge weights (i.e., the number of perturbation sampling times). The value range of the weight stability index is [0,1]. The closer SI_xy is to 1, the higher the stability of the associated edge in multiple perturbation tests, and the more reliable the physical transmission relationship it represents; the closer SI_xy is to 0, the more sensitive the associated edge is to perturbation, and the transmission relationship it represents may be a spurious correlation caused by noise or random factors. Through the above dynamic weight perturbation operation, the true physical transmission relationship can be effectively distinguished from noise and spurious correlations in the data, thereby improving the reliability of subsequent collaborative risk judgment.

[0045] In an optional embodiment, edges in the heterogeneous association graph structure whose weight stability index is greater than a preset stability threshold are marked as valid transmission edges. A set of stable transmission paths is determined based on the connectivity of these valid transmission edges in the heterogeneous association graph structure. Specifically, this includes: performing a stability comparison analysis between the weight stability index of the edges and the preset stability threshold to obtain a stability comparison result; marking edges whose weight stability index is greater than the preset stability threshold as valid transmission edges and marking edges whose weight stability index is less than or equal to the preset stability threshold as unstable edges; removing all unstable edges from the heterogeneous association graph structure to obtain an effective transmission subgraph, which retains the valid transmission edges and all nodes connected to them, including monitoring nodes and influencing factor nodes; and performing a connected component detection operation in the effective transmission subgraph to identify multiple independent connected sub-regions. Each of the multiple connected sub-regions includes a set of nodes interconnected by effective transmission edges. Using each influencing factor node in the node set of each connected sub-region as the path start node and each monitoring node in the node set of each connected sub-region as the path end node, a path search operation is performed to obtain candidate transmission paths from the path start node to the path end node. Each candidate transmission path includes the effective transmission edges and intermediate nodes traversed from the path start node to the path end node. The mean path stability is determined based on the weight stability index of the effective transmission edges in the candidate transmission paths. The mean path stability is compared with a preset path stability threshold to obtain the path stability comparison results. Based on the path stability comparison results, candidate transmission paths with a mean path stability greater than the preset path stability threshold are determined as stable transmission paths. A set of stable transmission paths is generated based on all stable transmission paths.

[0046] In the above embodiment, the preset stability threshold can be set to 0.8. The selection of this preset stability threshold is based on the following: when a certain associated edge remains stable in more than 80% of the perturbation tests (i.e., the weight stability index > 0.8), the physical transmission relationship represented by that edge can be considered reliable. Stability comparison analysis is performed according to the following conditions: associated edges with a weight stability index > 0.8 are marked as effective transmission edges; associated edges with a weight stability index ≤ 0.8 are marked as unstable edges. The effective transmission subgraph is constructed as follows: all unstable edges are removed from the original heterogeneous association graph structure, and all effective transmission edges and all nodes connected to these effective transmission edges (including monitoring nodes and influence factor nodes) are retained. After removing isolated nodes (i.e., nodes not connected to any effective transmission edges), the effective transmission subgraph is obtained. The connected component detection operation is implemented using a depth-first search (DFS) algorithm. The specific steps are as follows: Step 1) Initialization: Mark all nodes as unvisited; Step 2) Starting from any unvisited node, perform a DFS traversal, grouping all traversed nodes into the same connected component; Step 3) Repeat Step 2) until all nodes are visited; Step 4) Finally, multiple independent connected sub-regions are obtained. The path search operation is implemented using the breadth-first search (BFS) algorithm. For each connected sub-region: take each influencing factor node in the sub-region as the path starting node; take each monitoring node in the sub-region as the path ending node; execute the BFS algorithm to search for all possible paths from the starting node to the ending node; record the effective transmission edge sequence and intermediate node sequence of each candidate transmission path. The formula for calculating the mean path stability is S_path=Σ(SI_i) / m, where SI_i is the weight stability index of the i-th effective transmission edge on the candidate transmission path; m is the total number of effective transmission edges on the candidate transmission path. The preset path stability threshold can be set to 0.75 in this embodiment. Path stability comparison analysis is performed according to the following conditions: candidate transmission paths with a mean path stability > 0.75 are identified as stable transmission paths; candidate transmission paths with a mean path stability ≤ 0.75 are excluded. All stable transmission paths are aggregated to generate a stable transmission path set. Each path in this set has the following characteristics: it originates from an influencing factor node; all associated edges along the path are valid transmission edges; and the overall stability of the path is high. These stable transmission paths represent reliable physical channels for the transmission of external influencing events to monitoring parameters, and are the basis for subsequent collaborative risk assessment.

[0047] In an optional embodiment, when a target transmission path that meets the collaborative risk triggering condition exists in the stable transmission path set, it is determined that the monitoring nodes in the target transmission path are in a multi-parameter collaborative risk state, and risk response intervention operations are performed on the target transmission path according to the multi-parameter collaborative risk state. Specifically, this includes: obtaining the total number of monitoring nodes and the total number of influencing factor nodes included in each stable transmission path in the stable transmission path set; performing a first comparison analysis of the total number of monitoring nodes with a preset node number threshold, and performing a second comparison analysis of the total number of influencing factor nodes with a preset influencing factor number threshold, to obtain the comparison results; and, based on the comparison results, removing nodes with a total number of monitoring nodes greater than a preset threshold. A stable transmission path with a threshold number of points and a total number of influencing factor nodes greater than or equal to a preset threshold number of influencing factors is identified as the target transmission path that meets the conditions for triggering collaborative risks. Real-time monitoring data of each monitoring node in the target transmission path is acquired, and the real-time monitoring data is compared with a preset safe data range to obtain the safety comparison results. Based on the safety comparison results, the degree of data deviation of each monitoring node in the target transmission path is determined, and the degree of data deviation is compared with a preset deviation threshold to obtain the deviation comparison results. If, based on the deviation comparison results, it is determined that there are monitoring nodes in the target transmission path with a data deviation greater than the preset deviation threshold, then... Each monitoring node in the target transmission path is in a multi-parameter collaborative risk state. The path collaborative risk index is determined based on the degree of data deviation and the location information of each monitoring node in the target transmission path. The risk response level of the target transmission path is determined based on the path collaborative risk index and the types of impact events of the influencing factor nodes in the target transmission path. When the risk response level is the early warning observation risk level, a data acquisition cycle adjustment command is issued to the IoT sensor devices corresponding to each monitoring node in the target transmission path to adjust the current data acquisition cycle of the IoT sensor devices to a high-frequency data acquisition cycle. Early warning observation is then performed on the target transmission path based on the path information and the path collaborative risk index. Task: When the risk response level is encrypted monitoring risk level, issue encrypted acquisition commands to IoT sensor devices to adjust the current data acquisition cycle of IoT sensor devices to encrypted data acquisition cycle, and generate auxiliary monitoring point deployment information based on the monitoring parameter type and location information of each monitoring node in the target transmission path; deploy auxiliary monitoring devices in the associated area of ​​the target transmission path according to the auxiliary monitoring point deployment information to perform encrypted collaborative monitoring tasks on the target transmission path; when the risk response level is emergency intervention risk level, issue real-time acquisition commands to IoT sensor devices to adjust the data acquisition mode of IoT sensor devices to real-time continuous acquisition mode.Based on the degree of data deviation and the path topology of the target transmission path, key risk transmission nodes are identified. Risk blocking tasks are then performed on the target transmission path based on the monitoring parameter types and risk transmission directions of these key risk transmission nodes.

[0048] In the above embodiments, the preset node number threshold can be set to 3 (of course, it can also be set to 4, 5, 6, etc., which is not limited here). When the number of monitoring nodes in a stable transmission path exceeds 3, it indicates that the path involves the coordinated changes of multiple monitoring parameters and has the potential characteristics of systemic risk. The preset impact factor number threshold can be set to 1 in this embodiment. This threshold means that the starting point of the stable transmission path must be at least one impact factor node, that is, the risk transmission must have a clear external triggering source. The complete judgment logic of the coordinated risk triggering condition is: condition 1 is that the total number of monitoring nodes > 3; condition 2 is that the total number of impact factor nodes ≥ 1. When conditions 1 and 2 are met at the same time, the stable transmission path is determined as the target transmission path. The preset safety data range is determined according to the monitoring parameter type and engineering level, as shown in Table 2 below: Table 2 Preset Safety Data Range Table The formula for calculating the data deviation is D_i = |X_i - X_center| / (X_max - X_min) × 2, where X_i is the real-time monitoring data of monitoring node i; X_center is the center value of the safe data range, X_center = (X_max + X_min) / 2; X_max and X_min are the upper and lower limits of the safe data range, respectively. The value range of the data deviation is [0, +∞). When D_i = 1, it indicates that the data is exactly at the safe boundary, and when D_i > 1, it indicates that the data has exceeded the safe range. The preset deviation threshold is set to 0.6 in this embodiment. When the data deviation of any monitoring node in the target transmission path is > 0.6, all monitoring nodes in the path are determined to be in a multi-parameter collaborative risk state. The path coordination risk index R is calculated using the formula R = α × D_avg × S_path + β × E_max, where D_avg is the average deviation of data from all monitoring nodes in the target transmission path; S_path is the mean path stability; E_max is the highest event severity coefficient of the influencing factor nodes in the target transmission path; α and β are weighting coefficients, which can be α = 0.6 and β = 0.4 (the values ​​of α and β are not limited here). The event severity coefficient E is set as follows: light rain / general construction E = 0.3; moderate rain / important construction E = 0.5; heavy rain / critical construction E = 0.7; torrential rain / emergency construction E = 1.0.

[0049] In the above embodiments, the criteria for classifying risk response levels are shown in Table 3 below: Table 3 Risk Response Level Classification Standards Path coordination risk index R Risk Response Level 0<R≤0.4 Early warning observation risk level 0.4<R≤0.7 Encryption monitoring risk level R>0.7 Emergency intervention risk level Intervention actions corresponding to the early warning observation risk level: Adjust the data collection cycle of IoT sensor devices from a regular cycle (e.g., 2 hours / time) to a high-frequency cycle (e.g., 30 minutes / time); continuously monitor the evolution trend of the target transmission path and record changes in the path coordination risk index. Intervention actions corresponding to the encrypted monitoring risk level: Adjust the data collection cycle to an encrypted cycle (e.g., 15 minutes / time); add temporary monitoring points at gaps in the target transmission path coverage area, with the deployment condition being that auxiliary monitoring points are deployed at the midpoint between two adjacent monitoring nodes; deploy auxiliary monitoring points between the influencing factor node and the nearest monitoring node, with the monitoring parameter type of the auxiliary monitoring points consistent with that of the adjacent main monitoring points. Deploy auxiliary monitoring equipment: Install portable monitoring equipment according to the deployment information. Intervention actions corresponding to the emergency intervention risk level are as follows: The data acquisition mode is adjusted to real-time continuous acquisition; in the target transmission path, the monitoring node with the largest data deviation and located in the middle of the path is identified as the key risk transmission node; targeted measures are taken according to the monitoring parameter type of the key risk transmission node: if it is a water pressure-related node, emergency precipitation measures are initiated; if it is a stress-related node, stress monitoring of the relevant supports is intensified and the need for temporary supports is assessed; if it is a displacement-related node, deformation monitoring of the relevant area is intensified and the need for reinforcement measures is assessed. Through the above-mentioned graded response mechanism, the deep foundation pit multi-parameter intelligent monitoring system can take corresponding intervention measures according to the severity of the collaborative risks, maximizing the safety of deep foundation pit construction while ensuring controllable monitoring costs.

[0050] Through the embodiments of this application, a heterogeneous association graph structure integrating physical constraints is constructed, and the stability of the transmission relationship is verified through dynamic weight perturbation operation. This realizes multi-parameter risk identification and hierarchical response based on path collaborative analysis, and improves the reliability of multi-parameter collaborative early warning for deep foundation pits.

[0051] The following describes the multi-parameter intelligent monitoring system for deep foundation pits in the embodiments of this invention from the perspective of hardware processing. (See attached document.) Figure 2 , Figure 2 This is a schematic diagram of the physical device structure of a multi-parameter intelligent monitoring system for deep foundation pits in this application embodiment.

[0052] It should be noted that, Figure 2 The structure of the deep foundation pit multi-parameter intelligent monitoring system shown is merely an example and should not impose any limitations on the functionality and scope of use of the embodiments of the present invention.

[0053] like Figure 2As shown, the deep foundation pit multi-parameter intelligent monitoring system includes a central processing unit (CPU) 201, which can perform various appropriate actions and processes according to a program stored in read-only memory (ROM) 202 or a program loaded from storage section 208 into random access memory (RAM) 203, such as executing the methods described in the above embodiments. The RAM 203 also stores... It contains various programs and data required for system operation. CPU 201, ROM 202, and RAM 203 are interconnected via bus 204. Input / output (I / O) interface 205 is also connected to bus 204.

[0054] The following components are connected to I / O interface 205: input section 206 including audio input devices, push-button switches, etc.; output section 207 including a liquid crystal display (LCD) and audio output devices, indicator lights, etc.; storage section 208 including a hard disk, etc.; and communication section 209 including a network interface card such as a LAN (Local Area Network) card, modem, etc. Communication section 209 performs communication processing via a network such as the Internet. Drive 210 is also connected to I / O interface 205 as needed. Removable media 211, such as a disk, optical disk, magneto-optical disk, semiconductor memory, etc., are installed on drive 210 as needed so that computer programs read from them can be installed into storage section 208 as needed.

[0055] In particular, according to embodiments of the present invention, the processes described above with reference to the flowcharts can be implemented as computer software programs. For example, embodiments of the present invention include a computer program product comprising a computer program carried on a computer-readable medium, the computer program containing computer programs for performing the methods shown in the flowcharts. In such embodiments, the computer program can be downloaded and installed from a network via communication section 209, and / or installed from removable medium 211. When the computer program is executed by central processing unit (CPU) 201, it performs the various functions defined in the present invention.

[0056] It should be noted that specific examples of computer-readable storage media may include, but are not limited to: electrical connections having one or more wires, portable computer disks, hard disks, random access memory (RAM), read-only memory (ROM), erasable programmable read-only memory (EPROM), flash memory, optical fiber, portable compact disc read-only memory (CD-ROM), optical storage devices, magnetic storage devices, or any suitable combination thereof. In this invention, a computer-readable storage medium can be any tangible medium containing or storing a program that can be used by or in conjunction with an instruction execution system, apparatus, or device.

[0057] The flowcharts and block diagrams in the accompanying drawings illustrate the architecture, functionality, and operation of possible implementations of systems, methods, and computer program products according to various embodiments of the present invention. Each block in a flowchart or block diagram may represent a module, program segment, or portion of code, which contains one or more executable instructions for implementing a specified logical function. It should also be noted that in some alternative implementations, the functions indicated in the blocks may occur in a different order than those shown in the drawings.

[0058] Specifically, the deep foundation pit multi-parameter intelligent monitoring system of this embodiment includes a processor and a memory. The memory stores a computer program. When the computer program is executed by the processor, it implements the IoT-based deep foundation pit multi-parameter intelligent monitoring method provided in the above embodiment.

[0059] In another aspect, the present invention also provides a computer-readable storage medium, which may be included in the deep foundation pit multi-parameter intelligent monitoring system described in the above embodiments; or it may exist independently and not assembled into the deep foundation pit multi-parameter intelligent monitoring system. The storage medium carries one or more computer programs, which, when executed by a processor of the deep foundation pit multi-parameter intelligent monitoring system, enable the deep foundation pit multi-parameter intelligent monitoring system to implement the IoT-based deep foundation pit multi-parameter intelligent monitoring method provided in the above embodiments.

[0060] The above-described embodiments are only used to illustrate the technical solutions of this application, and are not intended to limit it. Although this application has been described in detail with reference to the foregoing embodiments, those skilled in the art should understand that modifications can still be made to the technical solutions described in the foregoing embodiments, or equivalent substitutions can be made to some of the technical features. Such modifications or substitutions do not cause the essence of the corresponding technical solutions to deviate from the scope of the technical solutions of the embodiments of this application.

[0061] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. This program can be stored in a computer-readable storage medium, and when executed, it can include the processes described in the above method embodiments. The aforementioned storage medium includes various media capable of storing program code, such as ROM or random access memory (RAM), magnetic disks, or optical disks.

Claims

1. A multi-parameter intelligent monitoring method for deep foundation pits based on the Internet of Things, characterized in that, include: Given the acquisition of multiple types of monitoring data for deep foundation pits, the cross-physical quantity transmission relationship between each monitoring parameter in the multiple types of monitoring data is determined, and a heterogeneous correlation graph structure is constructed based on the cross-physical quantity transmission relationship. The heterogeneous correlation graph structure includes monitoring nodes and influencing factor nodes. The monitoring nodes correspond one-to-one with each monitoring parameter, and the influencing factor nodes correspond one-to-one with the external influencing events of the deep foundation pit. The correlation edges included in the heterogeneous correlation graph structure represent the physical transmission paths between the monitoring nodes and between the monitoring nodes and the influencing factor nodes. Spatiotemporal transmission features are extracted from the multi-type monitoring data on the heterogeneous association graph structure to obtain the temporal change features of the monitoring nodes and the transmission influence features of the association edges. Dynamic weight perturbation operation is performed on the association edges according to the temporal change features and the transmission influence features to obtain the weight stability index of the association edges. In the heterogeneous association graph structure, the association edges whose weight stability index is greater than a preset stability threshold are marked as effective transmission edges, and a set of stable transmission paths is determined based on the connectivity of the effective transmission edges in the heterogeneous association graph structure. When there is a target transmission path in the set of stable transmission paths that meets the collaborative risk triggering condition, it is determined that the monitoring node in the target transmission path is in a multi-parameter collaborative risk state, and a risk response intervention operation is performed on the target transmission path according to the multi-parameter collaborative risk state. The collaborative risk triggering condition is that the total number of monitoring nodes in the target transmission path is greater than a preset node number threshold and the target transmission path starts with an influencing factor node.

2. The method according to claim 1, characterized in that, In the case of acquiring multi-type monitoring data of deep foundation pits, the method of determining the cross-physical quantity transmission relationship between various monitoring parameters in the multi-type monitoring data, and constructing a heterogeneous correlation graph structure based on the cross-physical quantity transmission relationship, specifically includes: The multi-type monitoring data is acquired, including the displacement monitoring data, the stress monitoring data, the water pressure monitoring data, and the tilt monitoring data; Obtain the distribution of the support layers, the connection relationship of the retaining walls, the influence range of the dewatering wells, and the monitoring location information of each monitoring parameter of the deep foundation pit; Physical quantity type matching analysis is performed on the multi-type monitoring data to determine the cross-physical quantity transmission relationship between different types of monitoring parameters; The structural force transmission path between the monitoring parameters is determined based on the distribution of the supporting layers, the connection relationship of the retaining walls, and the influence range of the dewatering wells; and the monitoring spatial distance between the monitoring parameters is determined based on the monitoring location information. The monitoring spatial distance is mapped to a preset distance influence function to obtain a distance weight coefficient that characterizes the degree of spatial proximity. A physical transmission constraint matrix is ​​generated based on the cross-physical quantity transmission relationship, the structural force transmission path, and the distance weighting coefficient. The element values ​​in the physical transmission constraint matrix represent the transmission strength coefficients between the different types of monitoring parameters. The heterogeneous correlation graph structure is constructed based on the physical transmission constraint matrix, the monitoring nodes, and the influencing factor nodes.

3. The method according to claim 2, characterized in that, The construction of the heterogeneous correlation graph structure based on the physical transmission constraint matrix, the monitoring nodes, and the influencing factor nodes specifically includes: Acquire the construction impact events and environmental impact events for each of the monitoring parameters; Create monitoring nodes for each monitoring parameter, create construction factor nodes for the construction impact event, and create environmental factor nodes for the environmental impact event; The conductivity intensity coefficient is compared with a preset conductivity threshold to identify two monitoring parameters with a valid conductivity relationship. The spatial range of construction impact is determined based on the event type and construction location of the construction impact event, and the spatial range of environmental impact is determined based on the event type and area of ​​effect of the environmental impact event. The construction impact space range is matched with the monitoring location information to obtain a first matching result, and the monitoring nodes whose monitoring location information is located within the construction impact space range are determined as construction-affected nodes based on the first matching result. The degree of construction transmission impact is determined based on the monitoring location information of the affected construction nodes and the monitoring location information of the corresponding construction factor nodes. The environmental impact spatial range is matched with the monitoring location information to obtain a second matching result. Based on the second matching result, the monitoring nodes whose monitoring location information is located within the environmental impact spatial range are identified as environmentally affected nodes. The degree of environmental transmission impact is determined based on the monitoring location information of the affected environmental nodes and the monitoring location information of the corresponding environmental factor nodes. The heterogeneous association graph structure is constructed based on the monitoring parameters, the construction-affected nodes, the degree of construction transmission impact, the environmentally affected nodes, and the environmentally affected nodes.

4. The method according to claim 1, characterized in that, The step of extracting spatiotemporal transmission features from the multi-type monitoring data on the heterogeneous association graph structure to obtain the temporal change features of the monitoring nodes and the transmission influence features of the association edges specifically includes: Acquire multiple time-series data segments of the monitoring node within a preset historical time window; The multiple time-series data segments are input into a time-series feature encoding network to obtain the time-series change characteristics of the monitoring nodes; From the temporal change features, a first change feature of the source node connected to the target associated edge and a second change feature of the target node connected to the target associated edge are determined, and the cross-correlation function of the source node and the target node is determined based on the first change feature and the second change feature; The temporal correlation coefficient between the source node and the target node is determined based on the cross-correlation function, and the propagation delay parameter is determined based on the peak time position of the cross-correlation function and the temporal correlation coefficient. The conduction activity index between the source node and the target node is determined based on the temporal correlation coefficient and the conduction delay parameter. The conduction activity index is then coupled with the target initial edge weight of the target associated edge to obtain the conduction response strength value of the associated edge. Obtain the first physical quantity type of the monitoring parameter corresponding to the source node and the second physical quantity type of the monitoring parameter corresponding to the target node, and query the cross-physical quantity conduction attenuation factor that is combined with the physical quantity type corresponding to the first physical quantity type and the second physical quantity type from the preset physical quantity type mapping table; The conduction response intensity value is modulated and calculated with the cross-physical quantity conduction attenuation factor to obtain the conduction influence characteristics.

5. The method according to claim 4, characterized in that, The step of performing a dynamic weight perturbation operation on the associated edges based on the temporal change characteristics and the transmission influence characteristics to obtain the weight stability index of the associated edges specifically includes: The first activity index of the source node is determined based on the time-series change characteristics, and the second activity index of the target node is determined based on the time-series change characteristics. The edge activation factor of the target associated edge is determined based on the first activity index and the second activity index, and the perturbation amplitude coefficient of the target associated edge is generated based on the edge activation factor and the transmission influence feature. Multiple sets of random disturbance vectors are generated according to a preset disturbance sampling strategy, and the multiple sets of random disturbance vectors are scaled with the disturbance amplitude coefficient to obtain multiple sets of target disturbance quantities. The multiple sets of target perturbation quantities are superimposed with the target initial edge weights to obtain multiple sets of perturbation edge weights, and the multiple sets of perturbation edge weights are updated to the heterogeneous association graph structure to determine the amount of graph topology connectivity change when the heterogeneous association graph structure experiences edge weight perturbation. The change in graph topology connectivity is compared with a preset connectivity sensitivity threshold to obtain the sensitivity comparison results. When the change in graph topology connectivity is determined to be greater than the connectivity sensitivity threshold based on the sensitivity comparison result, a sensitive connectivity label is assigned to the corresponding perturbation edge weight; when the change in graph topology connectivity is determined to be less than or equal to the connectivity sensitivity threshold based on the sensitivity comparison result, a stable connectivity label is assigned to the corresponding perturbation edge weight. The weight stability index of the target associated edge is determined based on the number of perturbation edge weights with the stable connectivity label among the multiple sets of perturbation edge weights.

6. The method according to claim 1, characterized in that, The step of marking the edges in the heterogeneous graph structure whose weight stability index is greater than a preset stability threshold as valid transmission edges, and determining a set of stable transmission paths based on the connectivity of the valid transmission edges in the heterogeneous graph structure, specifically includes: The stability index of the associated edge is compared with the preset stability threshold to identify the associated edge with the weight stability index greater than the preset stability threshold as an effective propagation edge, and the associated edge with the weight stability index less than or equal to the preset stability threshold as an unstable edge. Remove all unstable edges from the heterogeneous relational graph structure to obtain an effective transmission subgraph; A connected component detection operation is performed in the effective transmission subgraph to identify multiple independent connected sub-regions; Using each influencing factor node in the node set of each connected sub-region as the starting node of the path and each monitoring node in the node set of each connected sub-region as the ending node of the path, a path search operation is performed to obtain candidate transmission paths. Stable transmission paths are determined from the candidate transmission paths based on the weight stability index of the effective transmission edges in the candidate transmission paths, and the set of stable transmission paths is generated.

7. The method according to claim 6, characterized in that, When a target transmission path that meets the collaborative risk triggering conditions exists in the set of stable transmission paths, the monitoring node in the target transmission path is determined to be in a multi-parameter collaborative risk state, and a risk response intervention operation is performed on the target transmission path according to the multi-parameter collaborative risk state, specifically including: Obtain the total number of monitoring nodes and the total number of influencing factor nodes included in each stable transmission path in the set of stable transmission paths; The stable transmission path that has a total number of monitoring nodes greater than the preset node number threshold and a total number of influencing factor nodes greater than or equal to the preset influencing factor number threshold is determined as the target transmission path that satisfies the collaborative risk triggering condition. Acquire real-time monitoring data of each monitoring node in the target transmission path, and determine the degree of data deviation of each monitoring node in the target transmission path based on the real-time monitoring data; Based on the degree of data deviation, determine whether each monitoring node in the target transmission path is in the multi-parameter collaborative risk state; When it is determined that each monitoring node in the target transmission path is in the multi-parameter collaborative risk state, the path collaborative risk index is determined based on the degree of data deviation and the location information of each monitoring node in the target transmission path; If the target transmission path is determined to be at the early warning observation risk level based on the path coordination risk index, an early warning observation task is performed on the target transmission path. If the target transmission path is determined to be at the encrypted monitoring risk level based on the path collaboration risk index, an encrypted collaborative monitoring task is performed on the target transmission path. If the target transmission path is determined to be at an emergency intervention risk level based on the path coordination risk index, a risk blocking task is performed on the target transmission path.

8. A multi-parameter intelligent monitoring system for deep foundation pits, characterized in that, The deep foundation pit multi-parameter intelligent monitoring system includes: one or more processors and a memory; the memory is coupled to the one or more processors, the memory is used to store computer program code, the computer program code includes computer instructions, and the one or more processors call the computer instructions to cause the deep foundation pit multi-parameter intelligent monitoring system to perform the method as described in any one of claims 1-7.

9. A computer-readable storage medium comprising instructions, characterized in that, When the instruction is executed on the deep foundation pit multi-parameter intelligent monitoring system, the deep foundation pit multi-parameter intelligent monitoring system performs the method as described in any one of claims 1-7.

10. A computer program product, characterized in that, When the computer program product is run on the deep foundation pit multi-parameter intelligent monitoring system, the deep foundation pit multi-parameter intelligent monitoring system performs the method as described in any one of claims 1-7.

Images