An intelligent monitoring and early warning system for geological disaster risk suitable for complex terrain power grid corridor

Abstract

The application is suitable for the technical field of geological disaster early warning, and provides a geological disaster risk intelligent monitoring and early warning system suitable for complex terrain power grid corridors, which comprises: collecting multi-source monitoring data through deploying a distributed multi-parameter sensing array, connecting each sensor node and an edge computing node by using an ad hoc network communication network, uploading the data to a cloud platform after preprocessing by the edge computing node, the cloud platform being internally provided with a multi-agent collaborative early warning model composed of a tower-level monitoring intelligent agent, a section-level collaborative intelligent agent and a basin-level decision-making intelligent agent, performing risk pattern recognition and abnormality judgment based on the preprocessed data uploaded by the edge computing node and the evaluation results reported by the section-level collaborative intelligent agent, and generating a collaborative control instruction to be fed back to the edge computing node to adjust the working parameters of the sensor node. The application effectively improves the accuracy, real-time performance and reliability of the geological disaster monitoring and early warning of the complex terrain power grid corridors.

Description

Technical Field

[0001] This invention relates to the field of geological disaster early warning technology, and in particular to an intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors with complex terrain. Background Technology

[0002] As power grid infrastructure extends into areas with complex terrain, power grid corridors often traverse mountainous and hilly regions prone to geological disasters. Landslides, collapses, and debris flows pose a serious threat to the safe operation of transmission lines. Currently, monitoring and early warning of geological disasters in power grid corridors mainly rely on multi-sensor deployment and centralized analysis monitoring systems. Data collected by various sensors is aggregated to a central server, which then performs data fusion analysis to generate early warning information. This approach not only places extremely high demands on the reliability of the communication network, but also suffers from data transmission interruptions due to severe signal blockage in mountainous areas. Furthermore, the centralized processing model lacks effective coordination mechanisms between monitoring points. When multiple risk points simultaneously exhibit anomalies, dynamic allocation of monitoring resources is impossible, and upstream and downstream correlation analysis based on risk propagation paths is difficult. The lack of modeling capabilities for the spatiotemporal evolution of risks in complex terrain environments results in early warnings that fail to meet the accuracy, real-time performance, and reliability requirements of practical engineering projects. Summary of the Invention

[0003] This invention provides an intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain, which addresses the problem that the accuracy, real-time performance, and reliability of early warnings are insufficient to meet the needs of actual engineering projects.

[0004] This invention provides an intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain, comprising: Distributed multi-parameter sensor arrays, edge computing nodes, cloud platforms, self-organizing network communication networks, operation and maintenance terminals, and user terminals; The distributed multi-parameter sensor array is deployed on the towers and key geological risk sections of the power grid corridor, and is used to collect multi-source monitoring data. The edge computing node is located at the convergence point of the line segment and is connected to each sensor node in the distributed multi-parameter sensor array through the self-organizing network communication network. The edge computing node is used to receive and preprocess the multi-source monitoring data collected by the distributed multi-parameter sensor array and upload the preprocessed data to the cloud platform. The cloud platform connects the edge computing nodes, operation and maintenance terminals, and user terminals respectively. The cloud platform has a built-in multi-agent collaborative early warning model, which includes multiple tower-level monitoring agents deployed on each sensor node, multiple segment-level collaborative agents deployed on each edge computing node, and a watershed-level decision-making agent deployed on the cloud platform. Based on the preprocessed data uploaded by the edge computing nodes and the evaluation results reported by the segment-level collaborative agents, the cloud platform calls the multi-agent collaborative early warning model to perform risk pattern recognition and anomaly judgment, and generates collaborative control commands. The cloud platform sends the analysis results to the operation and maintenance terminal, user terminal, and edge computing node; enabling the operation and maintenance terminal to generate operation and maintenance instructions and feed them back to the operation and maintenance personnel, the user terminal to generate early warning notifications and feed them back to the user, and the edge computing node to adjust the working parameters of the corresponding sensor nodes in the distributed multi-parameter sensor array.

[0005] Furthermore, the distributed multi-parameter sensor array includes several sensor nodes, each independently equipped with a microseismic sensor, a global navigation satellite system displacement sensor, a tilt sensor, a rainfall sensor, and a soil moisture sensor; wherein: The microseismic sensor is used to collect weak vibration signals generated before rock mass fracturing and slope instability. The global navigation satellite system displacement sensor is used to collect three-dimensional displacement data of the Earth's surface at the location of the sensor node; The tilt sensor is used to collect data on the tilt angle changes of the tower and slope; The rain sensor is used to collect real-time rainfall intensity and cumulative rainfall data; The soil moisture sensor is used to collect data on the volumetric moisture content changes of the slope soil.

[0006] Furthermore, the edge computing node is used to receive and preprocess the multi-source monitoring data collected by the distributed multi-parameter sensor array, and upload the preprocessed data to the cloud platform, including: The received raw monitoring data is preprocessed, including spatiotemporal synchronization alignment, outlier removal, missing data interpolation and completion, and data compression encoding. The preprocessed data is fused from multiple sources to extract key feature vectors characterizing geological risks. These key feature vectors include deformation acceleration, microseismic energy accumulation, effective rainfall accumulation, and soil saturation change rate.

[0007] Furthermore, the multiple tower-level monitoring intelligent agents are used to make a preliminary local risk assessment based on the multi-source monitoring data collected by their respective sensor nodes, report the preliminary assessment results to the corresponding section-level collaborative intelligent agent, and receive collaborative control commands issued by the section-level collaborative intelligent agent and adjust the working parameters of their respective sensor nodes.

[0008] Furthermore, the preliminary assessment of local risk based on multi-source monitoring data collected by the sensor node includes: Acquire multi-source monitoring data collected by the sensor node, including microseismic signals, surface displacement data, tilt angle data, rainfall intensity data, and soil moisture content data; Based on the multi-source monitoring data, a preliminary local risk assessment index is calculated, which includes the deformation rate, frequency of microseismic events, rainfall intensity, and change in soil moisture content at the current moment. The local risk preliminary assessment indicators are compared with the preset local early warning thresholds to determine whether any indicator exceeds the corresponding local early warning threshold. When any indicator exceeds the corresponding local warning threshold, a local warning message is generated. The local warning message includes the type of indicator exceeding the limit, the magnitude of the exceeding limit, and the time of exceeding the limit.

[0009] Furthermore, the multiple segment-level collaborative intelligent agents are used to aggregate the preliminary judgment results reported by multiple pole-level monitoring intelligent agents within their respective line segments, conduct local risk assessments, generate collaborative control commands and issue them to each pole-level monitoring intelligent agent within their respective line segments, report the assessment results to the basin-level decision-making intelligent agent, and receive collaborative scheduling commands issued by the basin-level decision-making intelligent agent.

[0010] Furthermore, the step of performing local risk assessment and generating collaborative control commands to be issued to each tower-level monitoring agent within the relevant line segment includes: Receive and aggregate the preliminary judgment results reported by each pole-level monitoring intelligent agent within the line segment. The preliminary judgment results include the pole identification corresponding to each pole-level monitoring intelligent agent, local risk preliminary judgment indicators, and local early warning information. Based on the initial assessment results, a section-level comprehensive risk index is calculated. The section-level comprehensive risk index includes the number of high-risk towers in the section, the connectivity of the risk transmission path, and the gradient of the overall deformation field of the section. The section-level risk comprehensive index is compared with the preset section-level early warning threshold. When the section-level risk comprehensive index exceeds the corresponding section-level early warning threshold, section-level early warning information is generated. Based on the comprehensive risk indicators and early warning information at the section level, collaborative control instructions are generated for each tower-level monitoring intelligent entity within the corresponding line section, and then sent to each tower-level monitoring intelligent entity within the corresponding line section.

[0011] Furthermore, the basin-level decision-making agent is used to assess basin-level risks based on the evaluation results reported by all segment-level collaborative agents, generate collaborative scheduling instructions to be issued to relevant segment-level collaborative agents, and push early warning information to operation and maintenance terminals and user terminals.

[0012] Furthermore, the basin-wide risk assessment based on the evaluation results reported by all segment-level collaborative agents includes: Receive and aggregate the evaluation results reported by all segment-level collaborative intelligent agents, including the line segment identifier, segment-level comprehensive risk index and segment-level early warning information corresponding to each segment-level collaborative intelligent agent; The preprocessed data uploaded by the edge computing node is obtained. The preprocessed data includes global deformation field data, global microseismic event distribution data and global rainfall distribution data after being processed by the edge computing node. The aggregated assessment results are integrated with the acquired preprocessed data to construct a basin-wide risk situation map; Based on the aforementioned basin-level risk situation map, and combined with the historical disaster case database, risk trend extrapolation is performed to generate basin-level risk assessment results.

[0013] Furthermore, the process of generating a basin-wide risk assessment result based on the aforementioned basin-level risk situation map and combined with a historical disaster case database includes: The basin-level risk situation map is matched with the historical risk situation map in the historical disaster case database. The historical disaster case database stores the risk situation map before the occurrence of historical geological disasters and the corresponding disaster evolution results. When a historical case with a similarity exceeding a preset threshold is matched, the disaster evolution result corresponding to that historical case is used as the basis for current risk trend projection. When no similar historical cases are found, risk trend projection is performed based on the physical mechanism model; The predicted risk evolution direction is comprehensively evaluated with the current risk level of each line segment to generate the final risk assessment result.

[0014] As can be seen from the above technical solutions, the present invention has the following advantages: The system provided by this invention collects multi-source monitoring data by deploying a distributed multi-parameter sensor array, connects each sensor node with an edge computing node using a self-organizing network communication network, and uploads the data to a cloud platform after preprocessing by the edge computing node. The cloud platform has a built-in multi-agent collaborative early warning model consisting of a tower-level monitoring agent, a section-level collaborative agent, and a watershed-level decision-making agent. Based on the preprocessed data uploaded by the edge computing node and the evaluation results reported by the section-level collaborative agent, the system performs risk pattern recognition and anomaly judgment, and generates collaborative control commands to be fed back to the edge computing node to adjust the working parameters of the sensor node. This invention preprocesses raw monitoring data locally via edge computing nodes, and combines this with the multi-hop relay capability of ad hoc communication networks to effectively solve the problems of easy communication interruption and low data transmission reliability under complex terrain conditions. By constructing a three-tiered hierarchical multi-agent collaborative architecture at the pole, section, and basin levels, a complete collaborative link from single-point perception and local collaboration to global decision-making is achieved, enabling each monitoring node to dynamically adjust its operating parameters according to the risk situation. Through basin-level decision-making agents fusing preprocessed data uploaded from edge computing nodes with evaluation results reported by section-level collaborative agents, risk trend extrapolation is performed, enhancing the ability to model the spatiotemporal evolution of risks in complex terrain environments. This significantly improves the accuracy, real-time performance, and reliability of geological disaster monitoring and early warning in complex terrain power grid corridors. Attached Figure Description

[0015] Figure 1 This is an architecture diagram of an intelligent monitoring and early warning system for geological disaster risks applicable to power grid corridors with complex terrain, as described in this invention. Figure 2 This is a schematic diagram illustrating the working principle of the pole-level monitoring intelligent agent in this invention; Figure 3 This is a schematic diagram illustrating the working principle of the segment-level collaborative intelligent agent in this invention; Figure 4 This is a schematic diagram illustrating the working principle of the watershed-level decision-making agent in this invention. Detailed Implementation

[0016] The terms “first,” “second,” “third,” “fourth,” etc. (if present) in the specification and accompanying drawings of this application are used to distinguish similar objects and are not necessarily used to describe a specific order or sequence. It should be understood that such data can be interchanged where appropriate so that the embodiments of this application described herein can be implemented in orders other than those illustrated or described herein. Furthermore, the terms “comprising” and “corresponding to,” and any variations thereof, are intended to cover non-exclusive inclusion; for example, a process, method, system, product, or apparatus that comprises a series of steps or units is not necessarily limited to those steps or units explicitly listed, but may include other steps or units not explicitly listed or inherent to such processes, methods, products, or apparatus.

[0017] Example 1 Please see Figure 1 This application's system includes a distributed multi-parameter sensor array, edge computing nodes, a cloud platform, a self-organizing network communication network, operation and maintenance terminals, and user terminals. The distributed multi-parameter sensor array is deployed on the towers and key geological risk sections of the power grid corridor, and is used to collect multi-source monitoring data. The edge computing nodes are located at the convergence points of the line sections and communicate with each sensor node in the distributed multi-parameter sensor array through the self-organizing network communication network. The edge computing nodes are used to receive and preprocess the multi-source monitoring data collected by the distributed multi-parameter sensor array, and upload the preprocessed data to the cloud platform. The cloud platform connects the edge computing nodes, operation and maintenance terminals, and user terminals respectively. The cloud platform has a built-in multi-agent collaborative early warning model, and the multi-agent... The multi-agent collaborative early warning model includes multiple tower-level monitoring agents deployed at various sensor nodes, multiple segment-level collaborative agents deployed at various edge computing nodes, and a watershed-level decision-making agent deployed on the cloud platform. Based on the preprocessed data uploaded by the edge computing nodes and the evaluation results reported by the segment-level collaborative agents, the cloud platform calls the multi-agent collaborative early warning model to identify risk patterns and determine anomalies, and generates collaborative control commands. The cloud platform sends the analysis results to the operation and maintenance terminal, user terminal, and edge computing nodes, enabling the operation and maintenance terminal to generate operation and maintenance instructions and feed them back to the operation and maintenance personnel, the user terminal to generate early warning notifications and feed them back to the user, and the edge computing nodes to adjust the working parameters of the corresponding sensor nodes in the distributed multi-parameter sensor array.

[0018] The working principle and process of the system of this invention will be described in detail below, taking the intelligent monitoring scenario of geological disaster risk in a power grid corridor with complex terrain as an example: First, distributed multi-parameter sensor arrays are deployed at various poles and key geological risk sections along the power grid corridor. Sensor nodes are installed near the foundations of each pole and at key locations on the slope. These sensor nodes are interconnected via a self-organizing network using multi-hop relays to overcome the wireless signal obstruction caused by mountainous terrain, ensuring reliable transmission of monitoring data to edge computing nodes located at the convergence points of the line segments. After receiving the raw monitoring data uploaded by the sensor nodes within their respective line segments, the edge computing nodes perform data preprocessing. The preprocessed multi-source monitoring data is then uploaded to the cloud platform via the self-organizing network. The cloud platform incorporates a multi-agent collaborative early warning model, which consists of pole-level monitoring agents deployed at each sensor node, segment-level collaborative agents deployed at each edge computing node, and watershed-level decision-making agents deployed on the cloud platform.

[0019] During system operation, the tower-level monitoring agent performs preliminary local risk assessment based on multi-source monitoring data collected by the sensor node it is located on. It calculates local risk assessment indicators such as deformation rate, frequency of microseismic events, rainfall intensity, and soil moisture content change at the current moment. These indicators are compared with preset local early warning thresholds. When any indicator exceeds the corresponding threshold, local early warning information is generated, and the preliminary assessment results are reported to the corresponding section-level collaborative agent. The segment-level collaborative intelligent agent aggregates the preliminary judgment results reported by each tower-level monitoring intelligent agent within its line segment. Combining the terrain features and power grid topology of the line segment, it calculates segment-level comprehensive risk indicators such as the number of high-risk towers, the connectivity of risk transmission paths, and the overall deformation field gradient of the segment. The segment-level comprehensive risk indicators are compared with preset segment-level early warning thresholds to generate segment-level early warning information. Based on the segment-level comprehensive risk indicators, it generates collaborative control instructions for each tower-level monitoring intelligent agent within its line segment. For example, it instructs specific tower-level monitoring intelligent agents to increase the sampling frequency, switch to encrypted upload mode, or activate backup sensors. The collaborative control instructions are then sent to each tower-level monitoring intelligent agent, and the evaluation results are reported to the basin-level decision-making intelligent agent. The basin-level decision-making agent gathers the assessment results reported by all section-level collaborative agents and acquires preprocessed data uploaded by edge computing nodes. The assessment results and preprocessed data are then fused to construct a basin-level risk situation map, which characterizes the risk level distribution and risk evolution trend of each line segment in the entire power grid corridor. Combined with a historical disaster case database, risk trend extrapolation is performed to generate basin-level risk assessment results, including the overall risk level, a list of high-risk sections, and the expected risk evolution direction.

[0020] The cloud platform sends the analysis results to the operation and maintenance terminal, user terminal, and edge computing nodes. The operation and maintenance terminal generates operation and maintenance instructions based on the received risk level and location, and sends them back to the operation and maintenance personnel in the form of work orders, guiding them to conduct on-site verification and handling in high-risk areas. The user terminal generates early warning notifications based on the received risk level and sends them back to relevant users via push notifications. The edge computing nodes adjust the operating parameters of the corresponding sensor nodes within their jurisdiction based on the collaborative control instructions issued by the cloud platform. For example, they may increase the sampling frequency of sensor nodes at high-risk towers, switch the upload mode to encrypted upload mode, or activate backup sensors, thereby achieving dynamic allocation of monitoring resources.

[0021] In this embodiment, the distributed multi-parameter sensor array includes several sensor nodes. Each sensor node is independently equipped with a microseismic sensor, a global navigation satellite system displacement sensor, a tilt sensor, a rainfall sensor, and a soil moisture sensor. Among them: the microseismic sensor is used to collect weak vibration signals generated before rock mass fracturing and slope instability; the global navigation satellite system displacement sensor is used to collect three-dimensional displacement data of the ground surface at the location of the sensor node; the tilt sensor is used to collect data on the tilt angle changes of the tower and slope; the rainfall sensor is used to collect real-time rainfall intensity and cumulative rainfall data; and the soil moisture sensor is used to collect data on the volumetric water content changes of the slope soil.

[0022] Specifically, the microseismic sensor employs a broadband accelerometer, embedded in the bedrock beneath the tower foundation or in the soil and rock mass at key locations on the slope. When micro-fractures occur within the rock mass or stress adjustments occur before slope instability, the microseismic sensor can capture weak vibration signals with frequencies ranging from 0.5 Hz to 200 Hz and convert these signals into electrical signals for output. The GNSS displacement sensor uses a high-precision GNSS receiver module, installed at the top of the tower or in stable and potentially deformable areas of the slope. By receiving GNSS satellite signals, it performs differential positioning calculations and outputs real-time three-dimensional coordinate data of the sensor node's location, capturing subtle changes in surface displacement. The tilt sensor uses a microelectromechanical system (MEMS) dual-axis inclinometer, attached to the tower body and slope surface. When the tower tilts or the slope deforms, the tilt sensor senses changes in the direction of the gravitational field and outputs data on the change in tilt angle relative to the horizontal plane. The rain gauge uses a tipping bucket rain gauge, installed on the crossarm of a pole or an independent support, to collect real-time data on rainfall intensity and cumulative rainfall. The sampling frequency is matched to the needs of geological disaster monitoring, enabling it to capture short-duration heavy rainfall events. The soil moisture sensor uses the frequency domain reflectance method, buried at different depths in the soil on the slope. It obtains volumetric moisture content data by measuring changes in the soil's dielectric constant, reflecting the saturation state of the slope soil.

[0023] In this embodiment, the edge computing node is used to receive and preprocess multi-source monitoring data collected by the distributed multi-parameter sensor array, and upload the preprocessed data to the cloud platform, including the following: 1. Perform data preprocessing on the received raw monitoring data. Data preprocessing includes spatiotemporal synchronization alignment, outlier removal, missing data interpolation and completion, and data compression encoding. 2. The preprocessed data is fused from multiple sources to extract key feature vectors that characterize geological risks. These key feature vectors include deformation acceleration, microseismic energy accumulation, effective rainfall accumulation, and soil saturation change rate.

[0024] Specifically, edge computing nodes use a network time protocol to synchronize data across nodes, unifying monitoring data from different sources onto the same time base to form a spatiotemporally aligned multi-source dataset. An outlier detection method based on statistical distribution is employed; monitoring data points deviating from the normal range by more than three standard deviations are identified as outliers and removed. For missing data due to communication interruptions or momentary sensor malfunctions, a linear interpolation method is used to interpolate and complete the data based on valid data from adjacent time points, ensuring the continuity of the data sequence. After data cleaning, the edge computing nodes compress and encode the data, converting the original data into a smaller encoding format, reducing the amount of data uploaded to the cloud platform. Edge computing nodes perform multi-source fusion of cleaned microseismic data, GNS displacement data, tilt data, rainfall data, and soil moisture content data. Second-order difference operations are performed on the GNS displacement or tilt data to obtain deformation acceleration values, used to characterize the acceleration trend of deformation. The squared amplitude of the microseismic signal is integrated within a time window to obtain the cumulative microseismic energy, used to characterize the intensity of rock mass fracturing activity. Real-time rainfall intensity data is weighted and accumulated according to the time decay coefficient to obtain the effective cumulative rainfall considering the influence of previous rainfall, used to characterize the comprehensive impact of rainfall on slope stability. First-order difference operations are performed on the soil moisture content data to obtain the change in moisture content per unit time, used to characterize the dynamic changes in soil saturation state. The extracted key feature vectors are compressed and encoded before being uploaded to the cloud platform for further analysis by the watershed-level decision-making agent.

[0025] Example 2 Please see Figure 2 The working principle of the tower-level monitoring intelligent agent is explained below: In this embodiment, multiple tower-level monitoring agents are used to make preliminary local risk assessments based on multi-source monitoring data collected by their respective sensor nodes, report the preliminary assessment results to their respective section-level collaborative agents, and receive collaborative control commands issued by the section-level collaborative agents and adjust the working parameters of their respective sensor nodes.

[0026] Multiple pole-level monitoring agents are deployed one-to-one with sensor nodes in a distributed multi-parameter sensor array, with one pole-level monitoring agent running on each sensor node. The core function of the pole-level monitoring agent is to perform preliminary analysis of the raw monitoring data locally on the sensor node. Once the sensor node collects the raw monitoring data, the pole-level monitoring agent immediately initiates its local analysis process, completing preliminary judgments without waiting for instructions from upper-level nodes. For clearly abnormal data, it can quickly generate local early warning information. Simultaneously, the pole-level monitoring agent receives collaborative control commands from the section-level collaborative agent, dynamically adjusting its own operating parameters according to the commands to achieve on-demand configuration of monitoring resources.

[0027] This includes conducting a preliminary assessment of local risks based on multi-source monitoring data collected by the sensor node, including: 101. Acquire multi-source monitoring data collected by the sensor node, including microseismic signals, surface displacement data, tilt angle data, rainfall intensity data, and soil moisture content data; 102. Calculate the preliminary local risk assessment indicators based on multi-source monitoring data. The preliminary local risk assessment indicators include the deformation rate at the current moment, the frequency of microseismic events, the rainfall intensity, and the change in soil moisture content. 103. Compare the initial local risk assessment indicators with the preset local early warning thresholds to determine whether any indicator exceeds the corresponding local early warning threshold; 104. When any indicator exceeds the corresponding local warning threshold, a local warning message is generated. The local warning message includes the type of indicator exceeding the limit, the magnitude of the exceedance, and the time of the exceedance.

[0028] Specifically, after acquiring multi-source monitoring data collected by the sensor nodes at its location, the tower-level monitoring agent calculates various local risk preliminary assessment indicators. For the deformation rate, the tower-level monitoring agent acquires continuous position or angle data from the displacement or tilt sensors of the Global Navigation Satellite System, calculates the displacement or angle change relative to the previous moment using first-order difference operations, and then divides it by the sampling time interval to obtain the deformation rate. The calculation formula is as follows: in, Indicates the current time The rate of deformation; Indicates the current time The displacement or angle value; Indicates the previous moment The displacement or angle value; Indicates the sampling time interval.

[0029] To determine the frequency of microseismic events, the tower-level monitoring agent acquires the continuous vibration signal waveform collected by the microseismic sensors, uses the short-time energy method to detect valid microseismic events in the vibration signal, and counts the number of microseismic events detected per unit time to obtain the frequency of microseismic events. The calculation formula is as follows: in, Indicates within the time window Frequency of microseismic events within the area; Indicates within the time window The number of valid microseismic events detected within the facility; The inner part represents the statistical time window.

[0030] For rainfall intensity, the tower-level monitoring agent acquires real-time rainfall data collected by rain gauges, calculates the rainfall per unit time, and obtains the rainfall intensity using the following formula: in, Indicates the current time The intensity of rainfall; Indicates the sampling time interval The cumulative rainfall within the area.

[0031] For the change in soil moisture content, the tower-level monitoring agent acquires continuous moisture content data collected by soil moisture sensors, and calculates the change in moisture content from the previous moment to the current moment using first-order difference operations. The calculation formula is as follows: in, Indicates the current time The change in soil moisture content; Indicates the current time Soil moisture content value; Indicates the previous moment The soil moisture content value.

[0032] The local early warning threshold here is a pre-set judgment benchmark based on the statistical distribution of historical monitoring data, used to distinguish between normal fluctuations and abnormal precursors. Each local risk preliminary judgment indicator corresponds to an independent local early warning threshold, and each threshold is determined based on the historical data distribution characteristics of that indicator under normal operating conditions. For deformation rate, historical deformation rate data continuously monitored by the sensor node during the stable period are collected, and its average value is calculated. and standard deviation The average value plus three standard deviations is used as the local early warning threshold for deformation rate. The calculation formula is: When the actual deformation rate This indicates a significant acceleration in deformation. Regarding the frequency of microseismic events, the average number of microseismic events per unit time under historical steady-state conditions was statistically analyzed. Add twice the standard deviation to the mean. Local early warning threshold for the frequency of microseismic events The calculation formula is: When the frequency of actual microseismic events This indicates a significant increase in rock mass fracturing activity. For rainfall intensity, a basic threshold is set based on the geological conditions of the area; for example, a rainfall intensity threshold of 10 mm per hour is set for soil slopes, and 20 mm per hour for rock slopes. For soil moisture content variation, the amplitude of moisture content variation within the normal fluctuation range in historical data is statistically analyzed, and the maximum normal variation amplitude is selected. As a local early warning threshold, when the actual moisture content changes... or When this occurs, it indicates a drastic change in the soil's moisture content. If any indicator exceeds its corresponding local warning threshold, it signifies an anomaly in the physical process it represents, potentially indicating the formation of a geological hazard risk. The risk level is even higher when multiple indicators exceed their thresholds simultaneously.

[0033] Example 3 Please see Figure 3 The working principle of segment-level collaborative intelligent agents will be explained below: In this embodiment, multiple segment-level collaborative intelligent agents are used to aggregate the preliminary judgment results reported by multiple pole-level monitoring intelligent agents within their respective line segments, conduct local risk assessments, generate collaborative control commands and issue them to each pole-level monitoring intelligent agent within their respective line segments, report the assessment results to the watershed-level decision-making intelligent agent, and receive collaborative scheduling commands issued by the watershed-level decision-making intelligent agent.

[0034] Multiple segment-level collaborative intelligent agents are deployed one-to-one with edge computing nodes, with one segment-level collaborative intelligent agent running on each edge computing node. This deployment method enables each line segment to have independent local collaboration and decision-making capabilities, allowing for the aggregation and analysis of data reported by multiple pole-level monitoring intelligent agents within its segment at the edge, avoiding network congestion and processing delays caused by uploading all preliminary judgment results to the cloud platform. The core function of the segment-level collaborative intelligent agent is to achieve information collaboration and resource allocation among multiple poles within the line segment. When the segment-level collaborative intelligent agent receives the preliminary judgment results reported by each pole-level monitoring intelligent agent within its segment, it immediately initiates a local risk assessment process, combining the terrain features and power grid topology of the line segment to determine whether there is a possibility of local risk clustering or risk transmission. Simultaneously, the segment-level collaborative intelligent agent generates collaborative control commands for each pole-level monitoring intelligent agent based on the risk assessment results, enabling dynamic adjustment of monitoring resources, such as increasing the monitoring density of high-risk poles and reducing the sampling frequency of low-risk poles to save energy. The segment-level collaborative intelligent agent will also report the evaluation results to the basin-level decision-making intelligent agent and receive collaborative scheduling instructions issued by the basin-level decision-making intelligent agent, so as to realize the collaborative linkage between the upper and lower level intelligent agents.

[0035] The process of conducting local risk assessments and generating collaborative control commands to be sent to the monitoring agents at the tower level within the relevant line segment includes the following: 101. Receive and aggregate the preliminary judgment results reported by each pole-level monitoring intelligent agent within the line section. The preliminary judgment results include the pole identification corresponding to each pole-level monitoring intelligent agent, local risk preliminary judgment indicators, and local early warning information. 102. Calculate the section-level comprehensive risk index based on the initial judgment results of the convergence. The section-level comprehensive risk index includes the number of high-risk towers in the section, the connectivity of the risk transmission path, and the gradient of the overall deformation field of the section. 103. Compare the comprehensive risk index at the section level with the preset section-level early warning threshold. When the comprehensive risk index at the section level exceeds the corresponding section-level early warning threshold, generate a section-level early warning message. 104. Based on the section-level comprehensive risk indicators and section-level early warning information, generate collaborative control instructions for each pole-level monitoring intelligent body within the line section, and send them to each pole-level monitoring intelligent body within the line section.

[0036] Specifically, the segment-level collaborative intelligent agent receives preliminary assessment results reported by each tower-level monitoring intelligent agent within its assigned line segment via an internal bus or network communication interface. The assigned line segment refers to the power grid corridor section managed by the segment-level collaborative intelligent agent, consisting of three to five consecutive towers that are geographically adjacent and have upstream and downstream connections in the power grid topology. The preliminary assessment results reported by each tower-level monitoring intelligent agent include a tower identifier to uniquely identify the tower reporting the preliminary assessment result; local risk preliminary assessment indicators, namely, specific values ​​for deformation rate, microseismic event frequency, rainfall intensity, and soil moisture content changes; and local early warning information, namely, the type, magnitude, and time of exceeding limits. The segment-level collaborative intelligent agent aggregates and integrates all reported preliminary assessment results according to the tower identifier, forming a panoramic data of the risk status of each tower within the line segment.

[0037] The segment-level collaborative intelligent agent calculates the segment-level comprehensive risk index based on the converged preliminary judgment results. The number of high-risk towers within a segment refers to the number of towers within that segment whose local risk preliminary judgment index exceeds the corresponding local warning threshold. The calculation formula is as follows: in, Indicates the number of high-risk towers within the section; This represents the tower index, with values ​​ranging from 1 to... ; This indicates the total number of poles and towers within this line segment; For indicator functions, when the first The value is 1 if any local risk preliminary assessment indicator of a tower exceeds the corresponding local early warning threshold, otherwise the value is 0.

[0038] Risk transmission path connectivity is used to characterize whether there are geographical or topological connections between high-risk towers, allowing risks to potentially propagate downstream along valleys, slopes, or transmission lines. The calculation formula is: in, This represents the connectivity of the risk transmission path, with a value ranging from 0 to 1. The larger the value, the higher the probability of risk transmission. Indicates the index of the connection path between high-risk towers; This indicates the total number of connection paths between high-risk towers; Indicates the first The spatial distance between adjacent high-risk towers along the connecting path; This represents the preset maximum risk transmission distance threshold; if the distance exceeds this threshold, the risk is considered difficult to transmit.

[0039] The overall deformation field gradient of a section is used to characterize the spatial variation in surface deformation at the locations of each tower within that section of the line. It is obtained by calculating the spatial differences in the deformation rates of each tower. The calculation formula is: in, This represents the gradient of the overall deformation field of the section; Indicates the first The deformation rate of each tower; This represents the average deformation rate of all towers within the line segment; the greater the deformation field gradient, the more significant the deformation differences at different locations within the segment, and the higher the risk of overall slope instability.

[0040] The section-level early warning threshold is a judgment benchmark determined based on historical section risk event data. It is used to distinguish between normal local fluctuations and abnormal states that require the initiation of section-level coordinated response. Each section-level comprehensive risk index corresponds to an independent section-level early warning threshold. For the number of high-risk towers within a section, the maximum number of high-risk towers within the section under historical stable conditions is calculated, and this maximum value plus one is used as the section-level early warning threshold for the number of high-risk towers. ,when This indicates that multiple towers in the section are simultaneously experiencing anomalies. For the connectivity of the risk transmission path, a threshold is set based on the terrain features and power grid topology of the section. The value is 0.5. This indicates a spatial correlation among high-risk towers, suggesting that risk may be transmitted. For the overall deformation field gradient of the section, the average value of the deformation field gradient under historical steady-state conditions is statistically analyzed. and standard deviation The average value plus twice the standard deviation is used as the segment-level early warning threshold for the deformation field gradient. The calculation formula is: when This indicates a significant increase in the spatial variation of deformation within the section. When any section-level comprehensive risk indicator exceeds the corresponding section-level early warning threshold, the section-level collaborative agent generates section-level early warning information, including the type of exceeding limit indicator, the extent of exceeding limit, and the identification information of the line section.

[0041] Collaborative control commands are operation commands issued by segment-level collaborative intelligent agents to tower-level monitoring intelligent agents. These commands dynamically adjust the operating parameters of the sensor nodes where each tower-level monitoring intelligent agent resides, enabling on-demand configuration of monitoring resources. Each collaborative control command includes a target tower identifier to specify the tower-level monitoring intelligent agent receiving the command; a command type field, such as increasing the sampling frequency, switching to encrypted upload mode, or activating a backup sensor; and a command parameter field, for example, specifying a new sampling period in an instruction to increase the sampling frequency. The segment-level collaborative intelligent agent then distributes the generated collaborative control commands to each tower-level monitoring intelligent agent within its segment via an ad hoc network communication network, achieving dynamic allocation of monitoring resources.

[0042] Example 4 Please see Figure 4 The working principle of the watershed-level decision-making agent will be explained in detail below: In this embodiment, the watershed-level decision-making agent is used to assess watershed-level risks based on the evaluation results reported by all segment-level collaborative agents, generate collaborative scheduling instructions to be sent to relevant segment-level collaborative agents, and push early warning information to operation and maintenance terminals and user terminals.

[0043] Deployed on a cloud platform, the basin-level decision-making agent is the highest-level agent in the multi-agent collaborative early warning model. It is responsible for global risk perception, situational analysis, and decision-making scheduling across the entire power grid corridor coverage area. The core function of the basin-level decision-making agent is to aggregate the assessment results reported by all section-level collaborative agents and acquire preprocessed data uploaded from edge computing nodes. This data is then fused and analyzed to identify the spatial distribution patterns and temporal evolution trends of risks at the basin scale, generating collaborative scheduling instructions for multiple sections. Unlike the tower-level monitoring agent, which focuses on single-point risks, and section-level collaborative agents, which focus on local coordination, the basin-level decision-making agent possesses a global perspective, comprehensively considering risk correlations, meteorological conditions, terrain features, and historical disaster patterns across the entire basin, achieving a higher level of decision support. The basin-level decision-making agent also handles interaction with maintenance personnel and end users, pushing risk assessment results as early warning information to maintenance terminals and user terminals, supporting maintenance decisions and public early warning.

[0044] Among these, basin-wide risk assessment is conducted based on the evaluation results reported by all segment-level collaborative agents, including the following: 101. Receive and aggregate the evaluation results reported by all segment-level collaborative agents. The evaluation results include the line segment identifier, segment-level comprehensive risk index and segment-level early warning information corresponding to each segment-level collaborative agent. The basin-level decision-making agent receives assessment results reported by all segment-level collaborative agents through the message queue service of the cloud platform. Each assessment result includes a line segment identifier, used to uniquely identify the line segment from which the assessment result was reported; it includes segment-level comprehensive risk indicators, namely the number of high-risk towers within the segment, the connectivity of the risk transmission path, and the overall deformation field gradient of the segment; and it includes segment-level early warning information, namely the type of exceedance indicator, the exceedance magnitude, and the identification information of the line segment. The basin-level decision-making agent aggregates and integrates all reported assessment results according to the line segment identifier to form a panoramic data of the risk status of each segment within the entire power grid corridor.

[0045] 102. Obtain the preprocessed data uploaded by the edge computing node. The preprocessed data includes global deformation field data, global microseismic event distribution data, and global rainfall distribution data after being processed by the edge computing node. Global deformation field data is continuous deformation field distribution data formed by interpolating the deformation acceleration feature vectors uploaded by all edge computing nodes according to spatial coordinates, reflecting the surface deformation acceleration at various locations within the entire power grid corridor. Global microseismic event distribution data is spatial distribution data of microseismic events formed by aggregating the microseismic energy accumulation feature vectors uploaded by all edge computing nodes according to spatial coordinates, reflecting the spatial distribution density of rock mass fracturing activity within the entire power grid corridor. Global rainfall distribution data is spatial distribution data of rainfall formed by fusing the effective cumulative rainfall feature vectors uploaded by all edge computing nodes with meteorological radar data, reflecting the spatial distribution and non-uniformity of rainfall within the entire power grid corridor.

[0046] 103. Integrate the aggregated assessment results with the acquired preprocessed data to construct a basin-wide risk situation map; The fusion process establishes a unified two-dimensional grid coordinate system based on the spatial geographic coordinates of the power grid corridor, with a grid resolution of 50 meters by 50 meters. The line segment identifiers from the assessment results are mapped onto the grid cells covered by those segments, and the comprehensive risk index for each segment is assigned to the corresponding grid cell, forming a segment risk distribution layer. The global deformation field data, global microseismic event distribution data, and global rainfall distribution data from the preprocessed data are treated as independent feature layers, aligned according to spatial coordinates, and then superimposed onto the same grid coordinate system. The basin-level risk situation map is a comprehensive data object composed of the above multiple layers. Each grid cell contains information such as the risk level, deformation acceleration value, accumulated microseismic energy, effective accumulated rainfall, and the identifier of the corresponding line segment.

[0047] 104. Based on the basin-level risk situation map and combined with the historical disaster case database, risk trend extrapolation is carried out to generate basin-level risk assessment results.

[0048] 1. Perform similarity matching between the watershed-level risk situation map and the historical risk situation map in the historical disaster case database. The historical disaster case database stores the risk situation map before the occurrence of historical geological disasters and the corresponding disaster evolution results; 2. When a historical case with a similarity exceeding a preset threshold is matched, the disaster evolution result corresponding to that historical case is used as the basis for current risk trend projection; 3. When no similar historical cases are found, risk trend projection is performed based on the physical mechanism model; 4. The risk evolution direction obtained from the deduction is comprehensively evaluated with the current risk level of each line segment to generate the final risk assessment result.

[0049] Specifically, the historical disaster case database is a pre-built knowledge base that stores risk situation maps and corresponding disaster evolution results within specific time windows before the occurrence of historical geological disasters, including information such as disaster type, occurrence time, impact range, and degree of damage. Similarity matching employs a cosine similarity calculation method based on spatial grids. The current watershed-level risk situation map and each historical risk situation map in the case database are converted into feature vectors, and the cosine similarity between the two vectors is calculated using the following formula: in, The value represents the similarity, ranging from 0 to 1. The larger the value, the more similar the two risk situation maps are. This represents the total number of grid cells in the basin-level risk situation map; This indicates the current watershed-level risk situation map, showing the first... The risk characteristic value of each grid cell is obtained by weighted fusion of the grid cell's risk level, deformation acceleration value, microseismic energy accumulation, and effective rainfall accumulation. The historical risk situation map shows the first... The risk characteristic value of each grid cell. The preset threshold is set to 0.85. When the similarity exceeds this threshold, the current risk situation is considered to be highly similar to historical cases.

[0050] When a historical case with a similarity exceeding a preset threshold is matched, the disaster evolution results corresponding to that historical case are used as the basis for current risk trend prediction. The disaster evolution results record the risk evolution path of that historical case from its current risk state to the occurrence of the disaster, including the time series of risk level increases, the direction of expansion of high-risk areas, and the location and time of the disaster. The watershed-level decision-making agent uses the evolution results of this historical case as a reference for current risk trend prediction, forecasting the direction of risk evolution in the future. When no similar historical cases are matched, risk trend prediction is performed based on a physical mechanism model. The physical mechanism model includes a rainfall-deformation coupling model and a landslide propagation path prediction model. The rainfall-deformation coupling model, based on slope stability analysis theory, uses the current effective cumulative rainfall as input to calculate the changing trend of the slope safety factor. When the safety factor drops below a critical value, a landslide is predicted to occur. The landslide propagation path prediction model, based on digital elevation model and fluid dynamics theory, simulates the path and impact range of the landslide mass moving along the terrain after a landslide occurs. The watershed-level decision-making agent uses the deduction results of the physical mechanism model as the basis for predicting the current risk trend.

[0051] The risk evolution direction includes the expected expansion direction of high-risk areas, the line segments where the risk level is expected to rise, and the expected time window for the risk to reach its peak. The comprehensive assessment process overlays the current risk level with the derived evolution direction to predict the overall risk level over a future period. The final risk assessment result includes the overall risk level, divided into four levels: low risk, medium risk, high risk, and extremely high risk; a list of high-risk sections, listing the identification of line segments expected to enter a high-risk state and their expected risk levels; and the expected risk evolution direction, describing the movement direction and spread range of high-risk areas. The basin-level decision-making agent uses the generated risk assessment result as the basis for generating collaborative scheduling instructions, and simultaneously pushes early warning information to operation and maintenance terminals and user terminals.

[0052] It is understood that those skilled in the art can combine various implementation methods in the above embodiments under the guidance of the above examples to obtain technical solutions with multiple implementation methods.

[0053] The above description is only a preferred embodiment of the present invention and is not intended to limit the present invention. Any modifications, equivalent substitutions, and improvements made within the spirit and principles of the present invention should be included within the protection scope of the present invention.

Claims

1. A smart monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain, characterized in that, include: Distributed multi-parameter sensor arrays, edge computing nodes, cloud platforms, self-organizing network communication networks, operation and maintenance terminals, and user terminals; The distributed multi-parameter sensor array is deployed on the towers and key geological risk sections of the power grid corridor, and is used to collect multi-source monitoring data. The edge computing node is located at the convergence point of the line segment and is connected to each sensor node in the distributed multi-parameter sensor array through the self-organizing network communication network. The edge computing node is used to receive and preprocess the multi-source monitoring data collected by the distributed multi-parameter sensor array and upload the preprocessed data to the cloud platform. The cloud platform connects the edge computing nodes, operation and maintenance terminals, and user terminals respectively. The cloud platform has a built-in multi-agent collaborative early warning model, which includes multiple tower-level monitoring agents deployed on each sensor node, multiple segment-level collaborative agents deployed on each edge computing node, and a watershed-level decision-making agent deployed on the cloud platform. Based on the preprocessed data uploaded by the edge computing nodes and the evaluation results reported by the segment-level collaborative agents, the cloud platform calls the multi-agent collaborative early warning model to perform risk pattern recognition and anomaly judgment, and generates collaborative control commands. The cloud platform sends the analysis results to the operation and maintenance terminal, user terminal, and edge computing node; enabling the operation and maintenance terminal to generate operation and maintenance instructions and feed them back to the operation and maintenance personnel, the user terminal to generate early warning notifications and feed them back to the user, and the edge computing node to adjust the working parameters of the corresponding sensor nodes in the distributed multi-parameter sensor array.

2. The intelligent monitoring and early warning system for geological disaster risks applicable to power grid corridors with complex terrain as described in claim 1, characterized in that, The distributed multi-parameter sensor array includes several sensor nodes, each independently equipped with a microseismic sensor, a global navigation satellite system displacement sensor, a tilt sensor, a rainfall sensor, and a soil moisture sensor; wherein: The microseismic sensor is used to collect weak vibration signals generated before rock mass fracturing and slope instability. The global navigation satellite system displacement sensor is used to collect three-dimensional displacement data of the Earth's surface at the location of the sensor node; The tilt sensor is used to collect data on the tilt angle changes of the tower and slope; The rain sensor is used to collect real-time rainfall intensity and cumulative rainfall data; The soil moisture sensor is used to collect data on the volumetric moisture content changes of the slope soil.

3. The intelligent monitoring and early warning system for geological disaster risks applicable to power grid corridors with complex terrain as described in claim 1, characterized in that, The edge computing node is used to receive and preprocess the multi-source monitoring data collected by the distributed multi-parameter sensor array, and upload the preprocessed data to the cloud platform, including: The received raw monitoring data is preprocessed, including spatiotemporal synchronization alignment, outlier removal, missing data interpolation and completion, and data compression encoding. The preprocessed data is fused from multiple sources to extract key feature vectors characterizing geological risks. These key feature vectors include deformation acceleration, microseismic energy accumulation, effective rainfall accumulation, and soil saturation change rate.

4. The intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain, as described in claim 1, is characterized in that... The multiple tower-level monitoring agents are used to make preliminary local risk assessments based on multi-source monitoring data collected by their respective sensor nodes, report the preliminary assessment results to the corresponding section-level collaborative agent, and receive collaborative control commands issued by the section-level collaborative agent and adjust the working parameters of their respective sensor nodes.

5. The intelligent monitoring and early warning system for geological disaster risks applicable to power grid corridors with complex terrain as described in claim 4, wherein the preliminary assessment of local risk based on multi-source monitoring data collected by the sensor node includes: Acquire multi-source monitoring data collected by the sensor node, including microseismic signals, surface displacement data, tilt angle data, rainfall intensity data, and soil moisture content data; Based on the multi-source monitoring data, a preliminary local risk assessment index is calculated, which includes the deformation rate, frequency of microseismic events, rainfall intensity, and change in soil moisture content at the current moment. The local risk preliminary assessment indicators are compared with the preset local early warning thresholds to determine whether any indicator exceeds the corresponding local early warning threshold. When any indicator exceeds the corresponding local warning threshold, a local warning message is generated. The local warning message includes the type of indicator exceeding the limit, the magnitude of the exceeding limit, and the time of exceeding the limit.

6. The intelligent monitoring and early warning system for geological disaster risks applicable to power grid corridors with complex terrain according to claim 1, characterized in that, The multiple segment-level collaborative intelligent agents are used to aggregate the preliminary judgment results reported by multiple pole-level monitoring intelligent agents within their respective line segments, conduct local risk assessments, generate collaborative control commands and issue them to each pole-level monitoring intelligent agent within their respective line segments, report the assessment results to the basin-level decision-making intelligent agent, and receive collaborative scheduling commands issued by the basin-level decision-making intelligent agent.

7. The intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain, as described in claim 6, is characterized in that... The process of conducting local risk assessments and generating collaborative control commands to be sent to each tower-level monitoring agent within the relevant line segment includes: Receive and aggregate the preliminary judgment results reported by each pole-level monitoring intelligent agent within the line segment. The preliminary judgment results include the pole identification corresponding to each pole-level monitoring intelligent agent, local risk preliminary judgment indicators, and local early warning information. Based on the initial assessment results, a section-level comprehensive risk index is calculated. The section-level comprehensive risk index includes the number of high-risk towers in the section, the connectivity of the risk transmission path, and the gradient of the overall deformation field of the section. The section-level risk comprehensive index is compared with the preset section-level early warning threshold. When the section-level risk comprehensive index exceeds the corresponding section-level early warning threshold, section-level early warning information is generated. Based on the comprehensive risk indicators and early warning information at the section level, collaborative control instructions are generated for each tower-level monitoring intelligent entity within the corresponding line section, and then sent to each tower-level monitoring intelligent entity within the corresponding line section.

8. The intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain according to claim 1, characterized in that, The basin-level decision-making intelligent agent is used to assess basin-level risks based on the evaluation results reported by all segment-level collaborative intelligent agents, generate collaborative scheduling instructions and issue them to relevant segment-level collaborative intelligent agents, and push early warning information to operation and maintenance terminals and user terminals.

9. The intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain, as described in claim 8, is characterized in that... The process of basin-wide risk assessment based on evaluation results reported by all segment-level collaborative agents includes: Receive and aggregate the evaluation results reported by all segment-level collaborative intelligent agents, including the line segment identifier, segment-level comprehensive risk index and segment-level early warning information corresponding to each segment-level collaborative intelligent agent; The preprocessed data uploaded by the edge computing node is obtained. The preprocessed data includes global deformation field data, global microseismic event distribution data and global rainfall distribution data after being processed by the edge computing node. The aggregated assessment results are integrated with the acquired preprocessed data to construct a basin-wide risk situation map; Based on the aforementioned basin-level risk situation map, and combined with the historical disaster case database, risk trend extrapolation is performed to generate basin-level risk assessment results.

10. The intelligent monitoring and early warning system for geological disaster risks suitable for power grid corridors in complex terrain, as described in claim 9, is characterized in that... The process of generating basin-level risk assessment results based on the aforementioned basin-level risk situation map and combined with a historical disaster case database includes: The basin-level risk situation map is matched with the historical risk situation map in the historical disaster case database. The historical disaster case database stores the risk situation map before the occurrence of historical geological disasters and the corresponding disaster evolution results. When a historical case with a similarity exceeding a preset threshold is matched, the disaster evolution result corresponding to that historical case is used as the basis for current risk trend projection. When no similar historical cases are found, risk trend projection is performed based on the physical mechanism model; The predicted risk evolution direction is comprehensively evaluated with the current risk level of each line segment to generate the final risk assessment result.

Images