An integrated method and system for sensing sediment detention dam monitoring information

Abstract

The present invention relates to the technical field of check dam monitoring, and discloses a method and system for perceiving and integrating check dam monitoring and control information, including: collecting the monitoring data of the check dam, performing short-term risk prediction at the edge node according to the monitoring data, triggering local early warning based on the short-term risk prediction, and starting the response mechanism; establishing an adaptive cooperation mechanism among the edge nodes, identifying sensor anomalies and local disaster chains through the exchange and comparison of monitoring data among the nodes; the cloud integrates the monitoring data uploaded by the edge nodes, generates a global risk prediction model through dynamic data fitting, and conducts hierarchical emergency response according to the risk prediction results. The method of the present invention improves the safety management efficiency of the check dam, and provides a scientific support and an intelligent decision-making basis for flood control and fighting.

Description

Technical Field

[0001] The present invention relates to the technical field of check dams monitoring, and specifically to a method and system for integrating the perception of check dams monitoring and control information. Background Art

[0002] Check dams are important water conservancy facilities widely used in soil and water conservation and agricultural irrigation in mountainous areas. Their stable operation is directly related to regional ecological security and economic development. However, due to complex terrain, variable climate, and natural aging during long-term operation, check dams face various potential risks such as rain erosion, dam body landslides, and flood impacts. Especially during extreme weather and flood-prone seasons, the operation safety of check dams becomes the top priority in flood control and disaster relief work.

[0003] Existing check dam monitoring technologies mainly rely on online monitoring devices such as video surveillance, water level sensors, and rainfall sensors, and achieve the sharing and centralized analysis of monitoring data through Internet transmission technology. These technical means have played an important role in enhancing the perception ability of the operation status of check dams. However, traditional monitoring systems usually rely on cloud processing architectures, and their monitoring and early warning functions are highly dependent on network transmission. Once communication interruptions or delays occur, monitoring data cannot be uploaded to the cloud for processing in a timely manner, and the generation of early warning information may also lag behind, thus affecting the decision-making efficiency and emergency response ability of flood control work.

[0004] In recent years, with the development of emerging technologies such as the Internet of Things (IoT), edge computing, and artificial intelligence, new solutions have been provided for check dam monitoring and early warning systems. Among them, the application of edge computing technology has attracted particular attention. Edge computing performs data processing and analysis at the device end close to the data source, reducing the dependence on network transmission and significantly enhancing the real-time performance and the ability to respond to emergencies of the system. This technical architecture can not only maintain the independent operation of the system in case of communication interruptions, but also generate preliminary early warnings based on real-time monitoring data, providing a reliable basis for subsequent decision-making. Summary of the Invention

[0005] In view of the above existing problems, the present invention is proposed.

[0006] Therefore, the technical problem solved by the present invention is that the existing technology is highly dependent on network transmission, and monitoring data cannot be uploaded to the cloud for processing in a timely manner when communication is poor, resulting in deficiencies in real-time performance, reliability, and robustness.

[0007] To solve the above technical problem, the present invention provides the following technical solution: A method for integrating the perception of check dams monitoring and control information, including: collecting the monitoring data of check dams, performing short-term risk prediction at edge nodes according to the monitoring data, triggering local early warnings based on the short-term risk prediction, and starting a response mechanism;

[0008] Establish an adaptive cooperation mechanism among edge nodes. Through the exchange and comparison of monitoring data among nodes, identify sensor anomalies and local disaster chains;

[0009] The cloud integrates the monitoring data uploaded by edge nodes, generates a global wind measurement model through dynamic data fitting, and conducts hierarchical emergency responses according to the risk prediction results.

[0010] As a preferred solution of an integrated method for sensing silt dam monitoring information according to the present invention, wherein: the monitoring data includes environmental data, hydrological data, dam body data, video image data, and data around the silt dam area.

[0011] As a preferred solution of an integrated method for sensing silt dam monitoring information according to the present invention, wherein: the short-term risk prediction includes using the ARIMA model to establish a lightweight prediction model based on time series, training with the data of edge nodes in the cloud, and deploying the trained model to the corresponding edge nodes;

[0012] Edge nodes collect monitoring data in real time, and input the rainfall and water level data of the current time window into the lightweight prediction model to obtain the predicted values of rainfall and water level in the future short time. Update the input data every minute, re-predict the subsequent short-term data, and form a continuous prediction result;

[0013] Set rainfall thresholds, water level thresholds, and water level rising rate thresholds according to historical data. When the predicted values of the lightweight prediction model exceed the set thresholds, trigger a local warning, and synchronously report the risk level and prediction results to the cloud.

[0014] As a preferred solution of an integrated method for sensing silt dam monitoring information according to the present invention, wherein: the adaptive cooperation mechanism includes establishing a disaster chain model in the cloud, standardizing the real-time monitoring data collected by edge nodes, updating the states of various factors in the disaster chain, and realizing the collaborative identification of regional disaster chains;

[0015] Each edge node regularly sends core monitoring data and prediction results to adjacent edge nodes. After receiving the data, the adjacent edge nodes compare the timestamps. If the data is lagged or advanced, it is marked as a time anomaly and a secondary verification is triggered;

[0016] The secondary verification calculates the time deviation between the data of adjacent edge nodes by comparing the transmitted data with the local monitoring data. If the time deviation is greater than the set threshold, a compensation mechanism is triggered;

[0017] The compensation mechanism interpolates and compensates for missing and abnormal data, verifies the compensated data using the disaster chain model, and updates the status of each factor in the disaster chain with the verified compensated data;

[0018] An alarm is issued according to the data output in the disaster chain model. When the upstream edge node of the disaster chain triggers an alarm, a collaborative early warning signal is sent to the downstream edge nodes of the disaster chain to achieve collaborative early warning.

[0019] As a preferred solution of an information perception integration method for checking and monitoring a warping dam according to the present invention, wherein: establishing the disaster chain model in the cloud includes that the cloud determines the trigger factor, transmission factor, and termination factor of the disaster chain according to the monitoring data, and uses correlation analysis to determine the strongly correlated variables between the data:

[0020]

[0021] Among them, represents and the correlation between; represents the th monitoring data in the factor; represents the th monitoring data in the factor; represents the average value of the factor; represents the average value of the factor;

[0022] After determining the strongly correlated variables, a causality test is performed. Through the lag relationship of the time series, it is judged whether one variable has predictive ability for another variable:

[0023]

[0024] Among them, represents the target variable at the current time ; represents the constant term of the basic level of the target variable; represents the value of the independent variable at the th time lag period; represents the regression coefficient of; represents the lag order; represents the regression coefficient of; represents the th value of the target variable at the time lag period; The variables passing the causality test are used as the chain nodes, and the direction of the causal chain is confirmed according to the time lag relationship;

[0025] Use a Bayesian network to construct a disaster chain model, where nodes represent causal factors, directed edges represent causal relationships, define a conditional probability table for each node to describe the influence of the parent node state on the child node state, use Bayesian inference to update the conditional probability, and send the constructed disaster chain model to the edge nodes for application.

[0026] As a preferred solution of an integrated method for sensing sediment dam monitoring information according to the present invention, wherein: the collaborative identification of the regional disaster chain includes that the edge node obtains the monitoring values of the triggering factor, the transmission factor and the termination factor according to the real-time collected monitoring data. If the monitoring value of the triggering factor reaches the preset triggering factor threshold, the triggering factor is marked as the triggered state. Based on the state of the triggering factor, calculate the current state probability of the transmission factor through Bayesian inference:

[0027]

[0028] Among them, represents the probability that the transmission factor is activated under the condition that the triggering factor has been activated; represents the probability that the triggering factor and the transmission factor are both activated; represents the prior probability of the triggering factor ; Based on the state of the transmission factor, calculate the risk level of the termination factor through Bayesian inference;

[0029] Dynamically calculate the time delay of chain propagation according to the states of the triggering factor and the transmission factor;

[0030] Each node calculates the risk score of the current area based on the calculated state probability:

[0031]

[0032] Among them, represents the risk score; represents the total number of causal factors; represents the weight of the causal factor; represents the current state probability of the causal factor; According to the risk score value, determine the chain state;

[0033] If all factors in the chain are activated, it is determined that the chain is complete, triggering a regional early warning. The upstream node of the disaster chain sends a collaborative early warning signal to the downstream node of the disaster chain, and at the same time uploads the state to the cloud for global analysis.

[0034] As a preferred solution of the integrated method for sensing sediment dam monitoring information described in the present invention, wherein: the global risk prediction model includes that the cloud regularly obtains real-time monitoring data from each edge node, extracts key features through time series analysis, and obtains key feature indicators;

[0035] Using historical operation data as training samples, a global risk prediction model is established. The cloud updates the parameters of the global risk prediction model using the least squares method according to the real-time monitoring data, obtains the global risk score, and conducts hierarchical emergency responses based on the global risk score.

[0036] A sediment dam monitoring information sensing integration system adopting any of the methods described in the present invention, wherein: a short-term prediction module collects the monitoring data of the sediment dam, conducts short-term risk prediction at the edge node according to the monitoring data, triggers a local warning based on the short-term risk prediction, and starts a response mechanism; an edge collaboration module establishes an adaptive collaboration mechanism between edge nodes, and identifies sensor anomalies and local disaster chains through the exchange and comparison of monitoring data between nodes; a global prediction module integrates the monitoring data uploaded by the edge nodes by the cloud, generates a global risk prediction model through dynamic data fitting, and conducts hierarchical emergency responses according to the risk prediction results.

[0037] A computer device includes: a memory and a processor; the memory stores a computer program, including: when the processor executes the computer program, the steps of any of the methods described in the present invention are implemented.

[0038] A computer-readable storage medium stores a computer program thereon, including: when the computer program is executed by a processor, the steps of any of the methods described in the present invention are implemented.

[0039] The beneficial effects of the present invention: The present invention integrates multi-source monitoring data uploaded by edge nodes through the cloud, and uses dynamic data fitting and a global risk prediction model to achieve accurate assessment of the operation status of the sediment dam and hierarchical emergency responses. By combining dynamic risk prediction with multi-logical condition judgment, the present invention significantly improves the real-time sensing ability of complex disaster chains, can timely identify sudden risks and link regional warnings. At the same time, the dynamic update mechanism based on the model enhances the adaptability of the system to environmental changes, effectively reducing the misjudgment and missed judgment rates. The overall design improves the safety management efficiency of the sediment dam, providing scientific support and an intelligent decision-making basis for flood control and disaster relief. Description of the Drawings

[0040] To more clearly illustrate the technical solutions of the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the drawings in the following description are only some embodiments of the present invention. For those of ordinary skill in the art, without creative efforts, other drawings can be obtained based on these drawings.

[0041] Figure 1 This is the overall flowchart of an integrated method for sensing silt dam monitoring and control information provided by an embodiment of the present invention. Specific embodiments

[0042] To make the above objects, features, and advantages of the present invention more obvious and understandable, the following will describe the specific embodiments of the present invention in detail with reference to the drawings of the specification. Obviously, the described embodiments are part of the embodiments of the present invention, rather than all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those of ordinary skill in the art without creative efforts shall fall within the protection scope of the present invention.

[0043] Example 1, referring to Figure 1 , which is an embodiment of the present invention, provides an integrated method for sensing silt dam monitoring and control information, including:

[0044] S1: Collect the monitoring data of the silt dam, perform short-term risk prediction on the edge node according to the monitoring data, trigger a local warning based on the short-term risk prediction, and start the response mechanism.

[0045] Furthermore, in the silt dam monitoring system, the edge node collects and integrates the monitoring data through multi-source sensors. The monitoring data covers environmental data (such as rainfall intensity, wind speed), hydrological data (such as water level change, flow velocity), dam body data (such as stress, displacement), video image data (such as real-time monitoring of dam body cracks), and data around the silt dam area (such as upstream incoming water, downstream basin pressure). Each type of monitoring data is collected in real time through an independent sensor network, and unified data format conversion and time synchronization are performed at the edge node.

[0046] Specifically, environmental data collection obtains real-time environmental information through devices such as rain gauges and anemometers. The data units are, for example, mm / h or m / s. After being input into the system, dynamic trend analysis is carried out. Hydrological data is provided in real time by water level gauges and current meters. The sampling frequency is at the minute level, and the data is stored in a sequence form to capture the hydrological dynamic changes within the basin. The collection of dam body data is based on stress sensors and displacement measurement devices. Through a sensor array arranged at key parts of the dam body, the distribution of the internal stress field of the dam body, the crack propagation situation, and the surface displacement amplitude are monitored. The generated real-time data ensures a comprehensive grasp of the structural safety of the dam body. Video image data is collected by high-resolution cameras and combined with AI image analysis algorithms to be used for real-time monitoring of the external state of the dam body and rapid detection of abnormal conditions. The data around the check dam area is transmitted by upstream monitoring stations to dynamically capture the overall operation state of the basin.

[0047] Furthermore, a lightweight ARIMA prediction model based on time series is deployed at the edge nodes. The lightweight prediction model is trained by the cloud based on historical monitoring data and then distributed to each edge node for real-time short-term risk prediction. During actual operation, the edge nodes collect the rainfall and water level data of the current time window every minute, input the data into the prediction model, predict the change trends of rainfall and water level within the next 10 minutes, and update the input data every minute to form continuous prediction results, ensuring the dynamic control of short-term trends.

[0048] In short-term risk prediction, the system sets thresholds for rainfall, water level, and water level rising rate respectively based on historical data, which are the key conditions for triggering early warnings. The threshold setting is based on the analysis of historical disaster data and regional characteristics. For example, the rainfall threshold may be set at 50 mm / h. When the predicted value exceeds this value, it indicates that an extreme rainfall event may occur. The water level threshold refers to the safety warning line (such as 10 m), and the water level rising rate threshold is used to capture sudden flood phenomena. When the output value of the prediction model exceeds any threshold, the edge node immediately triggers a local early warning and reports the risk level and prediction results to the cloud through the communication module at the same time. The risk level is calculated by the edge node according to the amplitude of the predicted value exceeding the threshold, and is divided into low risk (close to the threshold), medium risk (exceeding the threshold by a certain range), and high risk (significantly exceeding the threshold).

[0049] While triggering local warnings, the edge nodes initiate a response mechanism, which specifically includes increasing the collection frequency of relevant monitoring data (such as increasing the original sampling frequency from once per minute to once every 10 seconds), synchronously enabling a backup sensor network to ensure redundant data collection, and sending risk signals to downstream associated nodes to make early emergency preparations. Meanwhile, the video surveillance data will enter a high-priority processing mode, and the system analyzes subtle changes in the video stream in real time through AI algorithms, such as the crack propagation rate or abnormal vibration of the dam body, to provide more detailed risk information. Through the local computing power of the edge nodes, the response mechanism can be triggered and implemented within an extremely short time, gaining more time for disaster prevention and mitigation.

[0050] S2: Establish an adaptive collaboration mechanism among edge nodes. By exchanging and comparing the monitoring data among nodes, identify sensor anomalies and local disaster chains.

[0051] The adaptive collaboration mechanism is a dynamic data interaction and analysis method based on distributed edge computing. Through real-time data sharing and collaborative processing among edge nodes, it improves the system's capabilities in anomaly detection, data correction, and regional disaster chain identification. In the monitoring scenario of check dams, the adaptive collaboration mechanism enables multiple edge nodes to form a dynamic and interdependent collaboration network, using the spatio-temporal correlation of data among nodes to make up for possible monitoring blind spots or errors of a single node.

[0052] Furthermore, establish a disaster chain model in the cloud. The edge nodes standardize the collected real-time monitoring data and update the status of each factor in the disaster chain to achieve collaborative identification of regional disaster chains.

[0053] The cloud establishes a disaster chain model based on the data of check dams. The cloud obtains key data for establishing the disaster chain model from the monitoring data and historical data, including rainfall, water level, dam body displacement, microseismic signals, etc. in the real-time data; flood and rainfall event records, dam body stress and deformation characteristics, etc. in the historical data; and geographical information data such as upstream and downstream watershed distributions, terrain slopes, and dam body materials.

[0054] The cloud determines the disaster trigger factor (such as rainfall), transfer factor (such as water level change), and termination factor (such as dam body landslide) based on the monitoring data, and uses correlation analysis to determine strongly correlated variables among the data:

[0055]

[0056] Among them, represents and the correlation between; represents the th monitoring data in the factor; Represents the th monitoring data in the factor; Represents the average value of the factor; Represents the average value of the factor.

[0057] According to the calculated correlation coefficient , if is greater than the set correlation threshold, for example , then it is considered that and have a strong correlation. After determining the strongly correlated variables, a causality test is performed. Through the lag relationship of the time series, it is judged whether one variable has predictive ability for another variable:

[0058]

[0059] Among them, represents the target variable at the current moment ; represents the constant term of the basic level of the target variable; represents the value of the independent variable at the th time lag period; represents 's regression coefficient; represents the lag order; represents 's regression coefficient; represents the value of the target variable at the th time lag period; The variables that pass the causality test are used as the chain nodes, and according to the time lag relationship, the direction of the causal chain is confirmed.

[0060] Use a Bayesian network to construct a disaster chain model. The nodes represent causal factors, the directed edges represent causal relationships, a conditional probability table is defined for each node to describe the influence of the parent node state on the child node state, and Bayesian inference is used to update the conditional probability. The constructed disaster chain model is sent to the edge nodes for application.

[0061] It should be noted that the key factors of the disaster chain include the triggering factor, the transmission factor, and the termination factor. Among them, the triggering factor is the starting point of the disaster chain, referring to the external environmental variables that trigger the disaster process, usually the driving factors that directly affect the downstream variables. The transmission factor is the intermediate link of the disaster chain, reflecting the degree of influence spread of the triggering factor and the trend of regional diffusion. The termination factor is the end link of the disaster chain, indicating the final impact of the disaster on the physical structure or ecological environment.

[0062] It should also be noted that the determination of the causal relationship chain needs to be established according to the actual environment. The monitoring environment of the warping dam has obvious regional characteristics, including topography, hydrological conditions, climate change, dam structure, and surrounding ecology, etc., all of which will affect the causal chain of disaster propagation. Specifically, the geographical environment and operating conditions of different warping dams are different. For example, rainfall has a more direct impact on the water level of warping dams in mountainous areas; warping dams in plain areas may be more affected by flow velocity and sediment accumulation.

[0063] There are differences in the types of disasters of warping dams, including but not limited to floods caused by rainfall, changes in dam body pressure caused by sediment accumulation, and dam body deformation caused by microseisms or landslides.

[0064] In actual application, the cloud will send the established disaster chain model to the edge nodes. The edge nodes will obtain the monitoring values of the triggering factor, transfer factor, and termination factor based on the real-time collected monitoring data. If the monitoring value of the triggering factor reaches the preset triggering factor threshold, the triggering factor will be marked as the triggered state. Based on the state of the triggering factor, the current state probability of the transfer factor will be calculated through Bayesian inference:

[0065]

[0066] Among them, represents the probability that the transfer factor is activated under the condition that the triggering factor has been activated; represents the probability that the triggering factor and the transfer factor are both activated; represents the prior probability of the triggering factor ; Based on the state of the transfer factor, the risk level of the termination factor is calculated through Bayesian inference.

[0067] According to the states of the triggering factor and the transfer factor, the time delay of chain propagation is dynamically calculated:

[0068]

[0069] Among them, represents the time delay of chain propagation; represents the delay time from the triggering factor to the transfer factor; represents the delay time from the transfer factor to the termination factor.

[0070] The delay time is analyzed by the cloud according to the real-time data, and the time difference between the change of the triggering factor (such as the rainfall reaching the threshold) and the response of the transfer factor (such as the significant increase in the water level change rate) in the real-time data is analyzed. The activation time of the triggering factor and the response time of the transfer factor are monitored, and the difference between the two times is the delay time from the triggering factor to the transfer factor.

[0071] Each node calculates the risk score of the current area based on the calculated state probability:

[0072]

[0073] Among them, represents the risk score; represents the total number of causal factors; represents the weight of the causal factor; represents the current state probability of the causal factor; According to the value of the risk score determine the state of the chain; If all factors in the chain are activated, it is determined that the chain is complete, trigger a regional warning, and the upstream node of the disaster chain sends a collaborative warning signal to the downstream node of the disaster chain, and at the same time upload the state to the cloud for global analysis.

[0074] Furthermore, each edge node regularly sends core monitoring data and prediction results to adjacent edge nodes. After receiving the data, the adjacent edge nodes compare the timestamps. If the data is lagged or advanced, it is marked as a time anomaly and a secondary verification is triggered.

[0075] Edge nodes regularly send the current core monitoring data and short-term risk prediction results to neighboring nodes through a low-latency communication protocol (such as MQTT or CoAP) to ensure data synchronization within the region. The core monitoring data includes real-time collected environmental data, dam body data, and the change trend of key factors in the short term of the prediction value. The comparison of timestamps is achieved through a highly accurate clock synchronized among nodes to ensure the timeliness and accuracy of the transmitted data. A lagged or advanced time anomaly may indicate communication delay, clock drift, or hardware failure between nodes.

[0076] The secondary verification calculates the time deviation between the data of adjacent edge nodes by comparing the transmitted data with the local monitoring data. If the time deviation is greater than the set threshold, a compensation mechanism is triggered. The compensation mechanism interpolates and compensates for missing and abnormal data, validates the compensated data using the disaster chain model, and updates the state of each factor in the disaster chain using the validated compensated data.

[0077] An alarm is issued according to the data output in the disaster chain model. When the upstream edge node of the disaster chain triggers an alarm, a collaborative warning signal is sent to the middle and downstream edge nodes of the disaster chain to achieve collaborative warning. By establishing a hierarchical linkage mechanism between the upstream and downstream edge nodes of the disaster chain, the downstream nodes can adjust their own monitoring strategies in real time according to the collaborative warning signal transmitted from the upstream to ensure unified perception and efficient response to risks within the region. The transmission of the collaborative warning signal adopts a redundant channel mechanism to ensure that the signal can still be transmitted through the backup path in case a single communication link fails.

[0078] S3: The cloud integrates the monitoring data uploaded by the edge nodes, generates a global risk prediction model through dynamic data fitting, and conducts hierarchical emergency responses according to the risk prediction results.

[0079] The cloud comprehensively integrates and analyzes the monitoring data transmitted from the edge nodes. Its design goal is to real-time judge the safety status of the environment and facilities where the check dams are located, and make corresponding hierarchical emergency response decisions according to the risk prediction results.

[0080] Furthermore, the cloud communicates with each edge node distributed on-site at regular intervals. These edge nodes not only monitor traditional basic environmental and structural indicators such as rainfall, water level, and dam body pressure, but also collect multi-dimensional information including electromagnetic interference, noise power density, common-mode noise voltage, vibration acceleration, and temperature gradient changes through various types of sensors. The cloud conducts unified conversion and standardization on the received multi-source heterogeneous data, and through predefined calibration and compensation strategies, eliminates outliers and fills in missing data to ensure the accuracy and consistency of the input data.

[0081] After ensuring the data quality, the cloud conducts feature extraction and dynamic data fitting on the processed data set. Referring to the established disaster chain model and regional characteristic parameters, the cloud extracts key feature indicators from long-term historical data and short-term real-time data through time series analysis methods and model-driven technologies. The time series analysis methods can include moving average models, long short-term memory networks, etc. The key features include (but are not limited to) load change rate, stress change rate, energy of equipment health status, dynamic trend of environmental interference power density, and steady-state and fluctuation characteristics of speed correction values.

[0082] The cloud realizes real-time update and fitting of risk scores by inputting the key feature indicators into a global risk prediction model, such as a Logistic regression model, a random forest model, or a support vector machine (SVM), so as to dynamically reflect the comprehensive operation risk of the check dam system during the continuous input of data.

[0083] After generating the preliminary risk score, the cloud does not directly give the final decision, but conducts further judgment and screening through a multi-logical condition determination module. For this purpose, the cloud has multiple sets of logical determination criteria, covering operation specifications in different dimensions and different time scales. For example, if the decline rate of the system load, the change rate of steam pressure, the status of equipment trip signals, and other key parameters within the past 30 minutes all meet the preset "planned operation" standard, the risk level will not be easily upgraded to a high risk. On the contrary, if any one of the logical conditions is not met or a sudden trip signal appears, the cloud will judge that the risk level has increased significantly and make corresponding adjustments to the risk level.

[0084] After the above multi-level dynamic assessment, the cloud finally forms a clear risk level classification result. When the system is judged to be low risk, the cloud maintains the regular monitoring and data analysis frequency without special intervention. When the system enters a medium-risk state, the cloud will automatically increase the frequency of data reporting, strengthen the monitoring density of specific subsystems, and require edge nodes to increase the frequency of correcting sensor data to ensure that potential problems can be discovered before the risk further expands. If the system enters a high-risk state, the cloud will immediately issue a coordinated early warning instruction to the downstream node, and closely monitor the data changes during the emergency process after the protective measures are implemented, and notify the management personnel or relevant decision-makers. This coordinated early warning signal not only requires downstream nodes to implement stricter monitoring strategies, but may also trigger automated protection devices (such as flood gate adjustment) and link the on-site emergency team to prepare for response.

[0085] This embodiment also provides an integrated system for monitoring and surveillance information perception of a silt dam, including a short-term prediction module, which collects monitoring data of the silt dam, performs short-term risk prediction at the edge node based on the monitoring data, triggers local warnings based on the short-term risk prediction, and starts a response mechanism; an edge collaboration module, which establishes an adaptive collaboration mechanism between edge nodes, identifies sensor anomalies and local disaster chains through the exchange and comparison of monitoring data between nodes; and a global prediction module, which integrates the monitoring data uploaded by the edge nodes in the cloud, generates a global risk prediction model through dynamic data fitting, and implements hierarchical emergency responses based on the risk prediction results.

[0086] Example 2, the following is an embodiment of the present invention, which provides a method for integrating information perception of a silt dam monitoring and supervision. In order to verify the beneficial effects of the present invention, scientific demonstration is carried out through economic benefit calculation and simulation experiments.

[0087] This embodiment is verified and optimized based on the field test conditions of the silt dam monitoring and early warning system to prove that through the adaptive coordination mechanism and short-term risk prediction model between edge nodes, regional disaster chains can be efficiently identified, abnormal data can be compensated, and the accuracy of early warning can be improved. The test environment selected a medium-sized mountain silt dam with complex terrain. The upstream catchment area is mostly valley terrain, and the downstream is a relatively flat irrigated farmland area. In order to restore the real working conditions, multiple edge nodes will be deployed around the silt dam: upstream nodes (denoted as A), midstream nodes (B), and downstream nodes (C).

[0088] During the experiment, A, B, and C collect environmental and hydrological data of the current time window from their respective sensors every minute, and use the lightweight ARIMA prediction model deployed on B to perform short-term predictions of rainfall and water level for the next 10 minutes. The model parameters were trained and optimized by the cloud based on historical monitoring data of the past year before the experiment, so as to ensure the adaptability of the model to the current season, climate conditions and the hydrological characteristics of the dam. As the core prediction node, B updates the prediction data every minute under normal conditions. If the predicted value exceeds the preset threshold (such as rainfall > 50mm / h or water level > 10m), B immediately triggers a local warning, increases the acquisition frequency of relevant sensors (from once a minute to once every 10 seconds), and reports the risk level and prediction results to the cloud and sends a risk signal to C. After receiving the collaborative warning signal from the upstream node, C can adjust its own monitoring strategy in advance, appropriately increase the monitoring density of sensitive areas or activate standby sensors to ensure that the downstream area has more time to respond to possible flood peaks or dam abnormalities.

[0089] To verify the adaptive cooperation mechanism between edge nodes, A and B, and B and C send each other core monitoring data and prediction results every 2 minutes. When adjacent nodes receive the data, they will compare them according to the timestamps of the data. If there is an obvious lag or lead (such as a deviation exceeding 100ms), it will be marked as a time anomaly and trigger a secondary verification. In this experiment, the rainfall, water level, and water level rising rate under different rainfall conditions will be analyzed, and the changes in data interaction, time deviation, and prediction accuracy before and after compensation between nodes will be recorded. By collecting multiple sets of experimental data, it can be proved that the method of the present invention can effectively improve data quality and warning accuracy, and compare the deficiencies of the existing single-node independent analysis mode. The experimental data obtained are analyzed, and a set of data for analysis is shown in Table 1.

[0090] Table 1 Experimental data table

[0091] Parameter Name A Data B Data C Data Time Difference (ms) Compensation Data Predicted Risk Level Rainfall (mm / h) 42 45 43 85 44 Low Risk Water Level (m) 9.5 9.7 9.6 90 9.65 Medium Risk Rate of Water Level Rise (m / h) 0.45 0.52 0.50 120 0.51 Medium Risk Data Deviation 0.02 0.04 0.03 110 0.03 Low Risk Comprehensive Risk Score 0.6 0.7 0.65 100 0.68 Medium Risk

[0092] A, B, and C in the tabular data form a complete local disaster chain. The compensated data reflects the unified result after interpolation compensation of abnormal or missing monitoring data under the node cooperation mechanism, and is the value after correcting the time deviation or data deviation existing in the original monitoring data between different nodes.

[0093] It should be noted that in actual applications, the compensated data is calculated by adjacent nodes of the edge node, and is not necessarily the data of the edge node of the disaster chain. In this embodiment, for the sake of simplified display, the average value is calculated through the data of the disaster chain nodes, which is only used to reflect the beneficial effect of the edge node cooperation mechanism.

[0094] As can be seen from the tabular data, the rainfall data of the upstream, midstream and downstream nodes are 42 mm / h, 45 mm / h and 43 mm / h respectively, indicating that there are certain differences between the monitoring points, but the deviation is not large. The existence of a time difference of 85 ms may cause data delay. After compensation, the rainfall is adjusted to 44 mm / h, and the weighted average of the three-point data is taken. This compensation mechanism ensures the overall consistency of the regional rainfall and avoids the interference of abnormal single-node data on system decision-making.

[0095] The water level monitored at the midstream node is 9.7 m, close to the warning line, and the water level rising rate of 0.52 m / h exceeds the threshold, indicating that the regional hydrological situation is in a high-risk state. After compensation, the water level and rate are adjusted to 9.65 m and 0.51 m / h respectively, which is consistent with the actual critical conditions. This adjustment avoids the possibility of misjudging high risk or low risk. Finally, the comprehensive risk score is 0.68, and the risk level is medium risk, providing a scientific and reasonable basis for emergency response for downstream nodes and management.

[0096] If the above functions are implemented in the form of software functional units and sold or used as independent products, they can be stored in a computer-readable storage medium. Based on this understanding, the technical solution of the present invention, in essence, or the part that contributes to the prior art, or a part of this technical solution, can be embodied in the form of a software product. This computer software product is stored in a storage medium and includes several instructions for causing a computer device (which can be a personal computer, a server, or a network device, etc.) to execute all or part of the steps of the methods described in various embodiments of the present invention. The foregoing storage medium includes: USB flash drives, mobile hard disks, read-only memories (ROM, Read-Only Memory), random access memories (RAM, Random Access Memory), magnetic disks, or optical disks and other various media that can store program codes.

[0097] The logic and / or steps represented in the flowchart or described in other ways herein, for example, can be considered as a definite sequence list of executable instructions for implementing logical functions, and can be specifically implemented in any computer-readable medium for use by an instruction execution system, apparatus, or device (such as a computer-based system, a system including a processor, or other systems that can fetch and execute instructions from the instruction execution system, apparatus, or device), or used in combination with these instruction execution systems, apparatus, or devices. For the purposes of this specification, a "computer-readable medium" can be any device that can contain, store, communicate, propagate, or transport a program for use by or in connection with an instruction execution system, apparatus, or device.

[0098] More specific examples (a non-exhaustive list) of computer-readable media include the following: an electrical connection (electronic device) having one or more wirings, a portable computer diskette (magnetic device), a random access memory (RAM), a read-only memory (ROM), an erasable programmable read-only memory (EPROM or flash memory), an optical fiber device, and a portable compact disc read-only memory (CDROM). Additionally, the computer-readable media can even be paper or other suitable media on which the program can be printed, since the program can be obtained electronically, for example, by optically scanning the paper or other media, followed by editing, interpretation, or otherwise processing it as appropriate, and then storing it in a computer memory.

[0099] It should be understood that the various parts of the present invention can be implemented by hardware, software, firmware, or a combination thereof. In the above-described embodiments, the multiple steps or methods can be implemented by software or firmware stored in a memory and executed by a suitable instruction execution system. For example, if implemented in hardware, as in another embodiment, any one or a combination of the following techniques well known in the art can be used: discrete logic circuits having logic gate circuits for implementing logical functions on data signals, application specific integrated circuits having appropriate combinational logic gate circuits, programmable gate arrays (PGAs), field programmable gate arrays (FPGAs), and the like.

[0100] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention and not to limit them. Although the present invention has been described in detail with reference to the preferred embodiments, those of ordinary skill in the art should understand that the technical solutions of the present invention can be modified or equivalently replaced without departing from the spirit and scope of the technical solutions of the present invention, and they should all be covered by the scope of the claims of the present invention.

Claims

1. A method for sensing and integrating silt dam monitoring and surveillance information, characterized in that: include: Collect monitoring data of the silt dam, make short-term risk predictions at edge nodes based on the monitoring data, trigger local warnings based on short-term risk predictions, and initiate response mechanisms; Establish an adaptive coordination mechanism between edge nodes to identify sensor anomalies and local disaster chains through the exchange and comparison of monitoring data between nodes; The adaptive collaborative mechanism includes establishing a disaster chain model in the cloud, where the edge nodes standardize the collected real-time monitoring data, update the status of each factor in the disaster chain, and realize the collaborative identification of regional disaster chains; Each edge node sends core monitoring data and prediction results to adjacent edge nodes at regular intervals. After receiving the data, the adjacent edge nodes compare the timestamps. If the data is delayed or advanced, it is marked as a time anomaly, triggering a secondary check. The secondary verification calculates the time deviation between the data of adjacent edge nodes by comparing the transmission data with the local monitoring data. If the time deviation is greater than the set threshold, the compensation mechanism is triggered; The compensation mechanism interpolates and compensates for missing and abnormal data, verifies the compensation data using the disaster chain model, and re-updates the status of each factor in the disaster chain using the verified compensation data; According to the data output in the disaster chain model, an alarm is issued. When the upstream edge node of the disaster chain triggers an alarm, a coordinated warning signal is sent to the downstream edge nodes of the disaster chain to achieve coordinated warning; The establishment of the disaster chain model in the cloud includes determining the triggering factor, transmission factor and termination factor of the disaster chain in the cloud according to the monitoring data, and using correlation analysis to determine the strongly correlated variables between the data: ； in, express and The correlation between express Factor Monitoring data; express Factor Monitoring data; express The mean of the factors; express The mean of the factors; After determining the strongly correlated variables, a causal test is performed to determine whether one variable has predictive power for another variable through the lag relationship of the time series: ； in, Indicates the current time The target variable; A constant term representing the base level of the target variable; Indicates The value of the independent variable at the time lag period; express The regression coefficient of represents the lag order; express The regression coefficient of Indicates The target variable value after the time lag period; the variables that pass the causal test are used as chain nodes, and the direction of the causal chain is confirmed according to the time lag relationship; Use Bayesian network to build disaster chain model, nodes represent causal factors, directed edges represent causal relationships, define conditional probability table for each node, describe the impact of parent node status on child node status, use Bayesian reasoning to update conditional probability, and send the built disaster chain model to edge nodes for application; The cloud integrates the monitoring data uploaded by the edge nodes, generates a global risk prediction model through dynamic data fitting, and implements hierarchical emergency response based on the risk prediction results.

2. A method for sensing and integrating silt dam monitoring and surveillance information according to claim 1, characterized in that: The monitoring data includes environmental data, hydrological data, dam body data, video image data and data around the silt dam area.

3. A method for sensing and integrating silt dam monitoring and surveillance information according to claim 2, characterized in that: The short-term risk prediction includes using the ARIMA model to establish a lightweight prediction model based on time series, using the data of the edge node for training in the cloud, and deploying the trained model to the corresponding edge node; The edge node collects monitoring data in real time and The rainfall and water level data are input into the lightweight prediction model to obtain the short-term rainfall and water level forecast values ​​in the future. The input data is updated every minute, and the subsequent short-term data is re-predicted to form a continuous prediction result. According to historical data, rainfall thresholds, water level thresholds and water level rise rate thresholds are set respectively. When the predicted value of the lightweight prediction model exceeds the set threshold, a local warning is triggered, and the risk level and prediction results are simultaneously reported to the cloud.

4. A method for sensing and integrating silt dam monitoring and surveillance information according to claim 3, characterized in that: The collaborative identification of the regional disaster chain includes that the edge node obtains the monitoring values ​​of the trigger factor, the transmission factor and the termination factor according to the real-time collected monitoring data. If the monitoring value of the trigger factor reaches the preset trigger factor threshold, the trigger factor is marked as a triggered state. Based on the state of the trigger factor, the current state probability of the transmission factor is calculated by Bayesian reasoning: ； in, Indicates the trigger factor Under conditions that have been activated, the transfer factor Probability of being activated; It represents the probability of simultaneous activation of the trigger factor and the transmission factor; Indicates trigger factor The prior probability of ; Based on the status of the transmission factor, the risk level of the termination factor is calculated by Bayesian reasoning; Dynamically calculate the time delay of chain propagation according to the status of trigger factors and transmission factors; Each node calculates the risk score of the current area based on the calculated state probability: ； in, represents the risk score; represents the total number of causal factors; represents the weight of the causal factor; Indicates the current state probability of the causal factor; based on the risk score The value of determines the chain status; If all factors in the chain are activated, the chain is judged to be complete, triggering a regional warning. The upstream node of the disaster chain sends a coordinated warning signal to the downstream node of the disaster chain, and uploads the status to the cloud for global analysis.

5. A method for sensing and integrating silt dam monitoring and surveillance information according to claim 4, characterized in that: The global risk prediction model includes: the cloud periodically obtains real-time monitoring data from each edge node, extracts key features through time series analysis, and obtains key feature indicators; The key characteristics include load change rate, stress change rate and equipment health status; Historical operating data is used as training samples to establish a global risk prediction model. The cloud uses least squares to update the parameters of the global risk prediction model based on real-time monitoring data, obtains the global risk score, and conducts hierarchical emergency response based on the global risk score.

6. A silt dam monitoring and surveillance information perception integration system, applied to a silt dam monitoring and surveillance information perception integration method according to any one of claims 1 to 5, characterized in that: include, The short-term prediction module collects monitoring data of the silt dam, performs short-term risk prediction at the edge node based on the monitoring data, triggers local warnings based on the short-term risk prediction, and starts the response mechanism; The edge collaboration module establishes an adaptive collaboration mechanism between edge nodes, and identifies sensor anomalies and local disaster chains through the exchange and comparison of monitoring data between nodes; The global prediction module integrates the monitoring data uploaded by the edge nodes in the cloud, generates a global risk prediction model through dynamic data fitting, and implements hierarchical emergency response based on the risk prediction results.

7. A computer device comprising: A memory and a processor; the memory stores a computer program, characterized in that: when the processor executes the computer program, the steps of a method for integrating silt dam monitoring and surveillance information perception are implemented as described in any one of claims 1-5.

8. A computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of a method for sensing and integrating silt dam monitoring and supervision information are implemented as described in any one of claims 1 to 5.

Images