Disaster event enhanced determination method and system based on monitoring node abnormal features

Abstract

The application discloses a kind of disaster event enhanced determination method and system based on monitoring node abnormal feature belongs to disaster monitoring and early warning technical field, this method includes: obtaining the event data and state data of multiple monitoring nodes in distributed monitoring network;According to abnormal determination rule, identify abnormal node;The spatial distribution characteristics of abnormal node are analyzed, to judge its spatial correlation with the event area detected;The time sequence relationship of abnormal occurrence time and event initial time and the consistency with event propagation order are analyzed;When spatial and time sequence characteristics meet preset condition simultaneously, generate enhanced criterion;The confidence or risk level of disaster event is corrected by fusing enhanced criterion and original disaster determination result.The application converts the abnormal data traditionally excluded into usable disaster feature information, improves the utilization rate of monitoring data and the identification ability of system to extreme disaster event, and the abnormal feature is only used as auxiliary criterion, to ensure the reliability of system.

Description

Technical Field

[0001] This invention belongs to the field of disaster monitoring and early warning technology, specifically relating to a method and system for enhancing the judgment of disaster events based on the abnormal behavior characteristics of monitoring nodes. Background Technology

[0002] Disaster monitoring and early warning systems typically rely on physical quantity data collected by nodes in a distributed monitoring network to determine disaster events. Traditional disaster determination methods mainly utilize the valid data reported by monitoring nodes for analysis. For example, in earthquake monitoring, the vibration intensity and arrival time information of seismic waveforms recorded by each station are used to locate the epicenter and estimate the magnitude. This type of method assumes that each monitoring node can function normally and output valid data, and draws conclusions about disaster events based on statistical analysis and model inference of the valid data.

[0003] However, during actual disasters, especially severe disasters, some nodes in the monitoring network may exhibit various anomalies due to the impact of the disaster itself. For example, a strong earthquake may cause signal interruptions, abnormal data fluctuations, or data loss at monitoring stations near the epicenter. Similarly, in industrial systems, severe malfunctions may cause some sensors to exhibit abnormal behaviors such as sudden data changes, signal distortion, or communication interruptions. These anomalies themselves contain important information about the intensity and scope of the disaster.

[0004] Existing technologies typically handle such abnormal data by directly discarding or marking it as invalid, relying solely on valid data output from normally functioning monitoring nodes for disaster assessment. While this approach ensures the quality of the data used, it has the following drawbacks: First, it loses the disaster characteristic information contained in the abnormal data, resulting in some important disaster intensity indicators not being effectively utilized; second, in severe disaster events, the large number of monitoring nodes malfunctioning significantly reduces the amount of usable valid data, potentially lowering the accuracy and reliability of disaster assessment; third, existing methods cannot utilize the behavioral patterns of abnormal nodes to reflect the spatial distribution and intensity characteristics of disasters, reducing the system's ability to identify extreme disaster events.

[0005] Therefore, it is necessary to propose a disaster event enhancement judgment method that can make full use of the abnormal features of monitoring nodes as auxiliary criteria, transforming abnormal data that is traditionally regarded as invalid into usable disaster feature information, thereby improving the comprehensiveness and accuracy of disaster event judgment. Summary of the Invention

[0006] To address the aforementioned shortcomings of existing technologies, the present invention aims to provide a disaster event enhancement judgment method and system based on the abnormal characteristics of monitoring nodes. This method systematically analyzes the abnormal behavior of monitoring nodes, extracting the spatial distribution and temporal characteristics of these abnormal nodes as enhancement criteria to assist in disaster event judgment, thereby improving the system's ability to identify severe disaster events.

[0007] To achieve the above objectives, the present invention adopts the following technical solution: A method for enhancing disaster event determination based on anomaly characteristics of monitoring nodes includes the following steps: Step S1, Data Acquisition and Basic Judgment: Acquire event data and status data from multiple monitoring nodes in the distributed monitoring network. Based on the event data, perform filtering and analysis to obtain preliminary disaster judgment results. The event data includes physical quantity data collected by each monitoring node, such as acceleration values, velocity values, and displacement values. The status data includes the operating status information (such as online status and power supply status) and data integrity information (such as whether the data sampling rate is normal and whether the data transmission is continuous) of each monitoring node. The preliminary disaster judgment results include the detected event area or event propagation path, the initial event time, and the original disaster judgment confidence level C0.

[0008] Step S2, Abnormal Node Identification: Analyze the status data of the multiple monitoring nodes and identify abnormal nodes according to preset abnormal judgment rules. The abnormal node is a node that meets at least one of the following conditions within a preset time window: no data output, that is, the duration of no continuous data output within the preset time window T exceeds the preset threshold T1; output data exceeds the preset normal range, that is, the output data value exceeds the preset multiple standard deviation range of the statistical distribution of the node's historical data; data changes abruptly or is interrupted, that is, the rate of change of adjacent sampling points exceeds the preset rate of change threshold.

[0009] Step S3, Abnormal Spatial Distribution Analysis: Obtain the geographical location information of the abnormal nodes and analyze their spatial distribution characteristics. Specifically, this includes: calculating the spatial distance between abnormal nodes to determine if they are spatially concentrated in a specific area; when the number of abnormal nodes within a preset radius R is not less than a preset number threshold N, it is determined that the abnormal nodes are concentrated in that specific area; further, it is determined that the specific area has a spatial correlation with the detected event area or event propagation path: calculate the spatial distance D between the center of the specific area and the center of the detected event; when D is less than a preset distance threshold D... max It can determine whether a spatial correlation exists; or whether a specific region is located within a preset buffer zone of the theoretical propagation path of a detected event.

[0010] Step S4, Anomaly Timing Analysis: Extract the anomaly occurrence time of the anomaly node and analyze the temporal relationship between the anomaly occurrence time and the event initiation time. Specifically, this includes extracting the timestamp t of the first occurrence of the anomaly at the anomaly node. a Obtain the initial start time t0 of the detected events. a The absolute value of the time difference between t0 and t0 is less than the preset time threshold. When an anomaly occurs, it is determined that the time of occurrence of the anomaly at the abnormal node is temporally correlated with the initial event time. Simultaneously, the consistency between the anomaly occurrence time and the event propagation order is analyzed. This analysis includes: sorting each anomaly node from its spatial distance to the center of the detected event in ascending order, and determining whether the timestamp of the first occurrence of the anomaly at each node monotonically increases or approximately monotonically increases with increasing distance. When the proportion of nodes satisfying the monotonically increasing condition exceeds a preset matching threshold, it is determined that the anomaly occurrence time and the event propagation order are consistent.

[0011] Step S5, Enhanced Criterion Generation: When the spatial distribution characteristics and temporal characteristics of the abnormal nodes meet preset conditions, based on the number of abnormal nodes N... a Abnormal area range S a Matching degree M with abnormal time series t The enhancement criterion value E is calculated according to the preset weighted fusion formula: E = w1×f(N a ) + w2×g(S a ) + w3×h(M t (1) In equation (1), E is the enhancement criterion value, and its value range is [0, 1]; N a The number of abnormal nodes required to satisfy the temporal correlation condition; S a M represents the area of ​​the abnormally concentrated region, in square kilometers; t The abnormal time sequence matching degree is the percentage of nodes whose abnormal occurrence time matches the event propagation order; w1, w2, and w3 are preset weight coefficients, corresponding to the weights of the number of abnormal nodes, the range of abnormal regions, and the abnormal time sequence matching degree, respectively, and satisfying w1 + w2 + w3 = 1; f(N a Let f(N) be a normalization function for the number of abnormal nodes, mapping the number of abnormal nodes to the interval [0, 1]. A linear normalization method can be used. a )=N a / N max , where N max To monitor the theoretically maximum number of nodes that can experience anomalies in the network; g(S a ) is a normalization function for the range of the outlier region, mapping the area of ​​the outlier region to the interval [0, 1]. This can be achieved using g(S). a )=Sa / S max , among which, S max To monitor the total area covered by the network; h(M) t ) is the normalization function for the abnormal time series matching degree, which can be directly expressed as the matching ratio h(M) t )=M t .

[0012] Step S6, Result Correction: Based on the obtained original disaster determination confidence level C0, if the original disaster determination confidence level C0 meets the preset initial determination threshold, an enhancement criterion E is introduced, and the original disaster determination result is corrected based on the enhancement criterion E to obtain the final disaster determination result; when the original disaster determination confidence level C0 is lower than the preset initial determination threshold, the enhancement criterion E is not used to correct the original disaster determination result; specifically, the expression for the corrected confidence level C is: (2) In equation (2), C is the corrected disaster event determination confidence level, with a value range of [0, 1]; C0 is the original disaster determination confidence level obtained based on normal event data analysis; The preset enhancement coefficient has a value range of [value range missing]. , is used to control the degree of contribution of the enhancement criterion to the final judgment result; E is the enhancement criterion value calculated by equation (1).

[0013] This formula ensures that the corrected confidence level C does not exceed 1, and when the original confidence level C0 is already high, the correction magnitude of the enhancement criterion is reduced accordingly.

[0014] Furthermore, in step S5, before calculating the enhancement criterion value E, an anomaly weight adjustment step is also included: weighting and reducing the contribution of each anomaly node to its anomaly characteristics based on the historical failure frequency of each anomaly node; specifically, calculating the reliability coefficient R of each anomaly node. i : R i = 1 - f i / f max (3) In equation (3), f i f is the historical failure frequency of the i-th node (in times / year). max For nodes with a high historical failure frequency, the reliability coefficient R is set to the preset maximum failure frequency threshold. i The smaller the value, the lower the weight of anomalous features; for nodes with a history of stable operation, the reliability coefficient R... i The larger the value, the higher the weight of the abnormal features; The normalization function for the number of abnormal nodes in formula (1) is f(N). aWhen ), the reliability coefficient R of each abnormal node is used. i We replace direct counting with weighted summation, i.e. The summation range is the range of abnormal nodes that satisfy the temporal correlation condition.

[0015] Furthermore, between steps S2 and S3, there is also an abnormality cause differentiation step: obtaining the communication link status information and environmental monitoring data of the abnormal node. When the communication link of the abnormal node is interrupted during the abnormality period and the environmental monitoring data of the area where the node is located is within the normal range, it is determined that the data abnormality of the abnormal node is caused by a communication failure. Here, a communication failure is manifested as a communication link interruption but the node is still collecting data normally locally, while a real abnormality caused by a disaster is manifested as an abnormality in the data collected locally by the node itself. The abnormal node is excluded from the abnormal nodes identified in step S2 so that it does not participate in the analysis of step S3 and subsequent steps.

[0016] This invention also provides a disaster event enhancement judgment system based on the abnormal characteristics of monitoring nodes, comprising: a data acquisition module, an anomaly identification module, a spatial analysis module, a time series analysis module, a criterion generation module, and a correction module. The functions of each module are consistent with the corresponding steps in the above methods. The modules communicate with each other through standardized data interfaces. The overall system can be deployed on a centralized server platform or a distributed computing platform, supporting real-time online operation.

[0017] Compared with the prior art, the present invention has the following beneficial effects: 1. This invention transforms abnormal monitoring node data, which is traditionally considered invalid, into usable disaster characteristic information, thereby improving the information utilization rate of monitoring data and avoiding information loss caused by simply removing abnormal data.

[0018] 2. By performing spatial distribution analysis and temporal analysis on abnormal nodes, this invention can effectively identify abnormal feature patterns in severe disaster events, thereby improving the system's sensitivity and identification capabilities for extreme disaster events.

[0019] 3. When the original disaster judgment result meets the preset initial judgment conditions, the present invention uses the abnormal features of abnormal nodes to form an enhanced criterion to correct the original judgment result, thereby improving the ability to identify extreme disasters while reducing the risk of misjudgment caused by equipment failure.

[0020] 4. The enhanced judgment method proposed in this invention is compatible with existing disaster judgment methods and can be integrated without changing the existing judgment process, thus having good engineering application value. Attached Figure Description

[0021] Figure 1This is a flowchart illustrating the disaster event enhancement determination method based on abnormal features of monitoring nodes according to the present invention.

[0022] Figure 2 This is a schematic diagram illustrating the correlation analysis between the spatial distribution of abnormal nodes and event regions according to the present invention.

[0023] Figure 3 This is a structural block diagram of the disaster event enhancement determination system of the present invention. Detailed Implementation

[0024] The present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments. The following embodiments are only used to illustrate the present invention and are not intended to limit the scope of the present invention.

[0025] Example 1: This embodiment uses an earthquake early warning scenario as an example to illustrate the specific implementation process of the disaster event enhancement judgment method based on the abnormal characteristics of monitoring nodes provided by this invention. Figure 1 As shown, the method includes six steps, S1 to S6.

[0026] Suppose an earthquake monitoring network has 200 monitoring stations deployed across a certain area. When an earthquake event occurs, the system executes the following steps: Step S1, Data Acquisition and Basic Judgment: The system acquires seismic waveform data and station operation status data reported by 200 monitoring stations in real time. Event data includes three-component acceleration waveform data recorded by each station, with a sampling rate of 100Hz. Status data includes the online status, power supply status, communication link status, and data sampling rate integrity information of each station.

[0027] Meanwhile, the system performs preliminary screening and analysis on the acquired data, identifying normal working nodes that have reported valid waveform data (assuming that there are 185 stations with normal data in this round). Based on the data of these normal working nodes, the system uses conventional seismic location algorithms (such as Geiger's method) to calculate preliminary disaster assessment results, including the detected event area (i.e., epicenter location O), the event initiation time t0, and the original disaster assessment confidence level C0 calculated based on the number of triggering stations and the signal-to-noise ratio. This preliminary disaster assessment result will serve as the basic input for subsequent spatial correlation analysis, temporal analysis, and result correction of abnormal nodes.

[0028] Step S2, Abnormal Node Identification: The system checks the status data of 200 monitoring stations one by one, setting a preset time window T as a 60-second period after the event occurs. Within this time window, the system determines abnormal nodes according to the following rules: 1. Signal interruption anomaly: If a station has no data output for more than 5 seconds within a time window T, it is determined to be a signal interruption anomaly node. For example, if station A does not report any data during the complete period from the 3rd second to the 60th second after the earthquake, it meets this condition.

[0029] 2. Data Anomaly Type: The system analyzes historical data from each station for the 30 days prior to the event and calculates the average output data from each station. and standard deviation If the output data value of a certain station exceeds the time window T, If the range is not specified, it is determined to be an abnormal node of the data anomaly type. For example, station B outputs an abnormal value that is more than 10 times the standard deviation of historical data after the earthquake.

[0030] 3. Abrupt anomaly: If the rate of change of adjacent sampling points of a station within the time window T exceeds the preset rate of change threshold, it is determined to be an abrupt anomaly node. For example, the data value jumps from the normal background noise level to the upper limit of the instrument range between two adjacent sampling points and then returns to zero.

[0031] Based on the above analysis, a total of 15 abnormal nodes were identified in this embodiment.

[0032] Step S3, anomaly spatial distribution analysis: such as Figure 2 As shown, the system obtains the geographic coordinates of 15 abnormal nodes, calculates the spatial distance between each abnormal node, sets a preset radius R=30 kilometers, and a preset quantity threshold N=5. The calculation shows that there are 12 abnormal nodes (including station A) in a circular area with a radius of 30 kilometers centered on station A, which meets the condition of concentrated distribution. The geometric center of this area is taken as the center point P of the abnormal concentrated area.

[0033] Further calculate the spatial distance D between the center P of the anomaly concentration area and the epicenter O of the detected earthquake event, and set a preset distance threshold D. max =50 km, calculated, D=18 km, which is less than D max The results indicate that the areas of concentrated anomalies are spatially correlated with the areas of earthquake events.

[0034] Step S4, Anomaly Time Sequence Analysis: The system extracts the timestamp t of the first occurrence of anomalies in 12 spatially correlated anomaly nodes. a The initial time t0 of the detected seismic event is obtained from data analysis of normal stations, and a preset time threshold is set. Seconds, analysis showed that 10 out of the 12 abnormal nodes had t values. a The absolute value of the time difference between t0 and t0 is less than 10 seconds, which satisfies the time sequence correlation condition.

[0035] Further analysis revealed the consistency between the timing of the anomaly and the sequence of event propagation: Station A, the closest to the epicenter O, was the first to exhibit the anomaly (t). a = t0+0.5 seconds), followed by station C, which is slightly farther away (t a = t0+2.1 seconds), and so on. The spatial distribution of the anomaly occurrence time is consistent with the pattern of seismic wave propagation from the epicenter outward, and the time series matching degree M t Relatively high.

[0036] Step S5, Enhanced Criterion Generation: Since the spatial distribution and temporal characteristics of the abnormal nodes both meet the preset conditions, the system generates enhanced criteria. The values ​​of each parameter are as follows: Number of abnormal nodes N a =10, normalization function f(N) a Linear normalization is used, N max Let it be 50, therefore f(N) a =10 / 50 = 0.2.

[0037] Abnormal area range S a The area of ​​the abnormally concentrated region. , where R actual In this embodiment, R represents the equivalent radius of the actual distribution of abnormal nodes. actual =25 km, S a Approximately 1963 square kilometers, normalization function g(S) a )=S a / S max S max Let's assume it's 50,000 square kilometers, therefore g(S) a =1963 / 50000≈0.039.

[0038] Abnormal timing matching degree M t Among the 12 spatially correlated anomaly nodes, the anomaly occurrence times of 10 nodes satisfy the condition of approximately monotonically increasing, M. t =10 / 12≈0.833, normalization function h(M) t )= M t =0.833.

[0039] With weighting coefficients w1=0.4, w2=0.2, and w3=0.4, and substituting them into equation (1), the enhanced criterion value E is calculated as follows: E = 0.4×0.2 + 0.2×0.039 + 0.4×0.833 = 0.08 + 0.008 + 0.333 = 0.421 Step S6, Result Correction: Assuming the original disaster assessment confidence level C0 = 0.75 based on normal station data, an enhancement coefficient is set. Substituting into equation (2), the corrected confidence level C is calculated as follows: C = 0.75 + 0.6×0.421×(1 - 0.75) = 0.75 + 0.063 = 0.813 The revised disaster event assessment confidence level has increased from 0.75 to 0.813. Simultaneously, the system uses a table mapping confidence level to risk level to determine whether an upgrade in risk level is necessary. For example, if a confidence threshold of 0.80 corresponds to a "high risk level," the revised assessment result will trigger an upgrade in risk level from "medium risk" to "high risk."

[0040] In this embodiment, anomalous features are only used as auxiliary criteria in the judgment. If the original confidence level C0 of the normal data analysis is lower than the initial judgment threshold (e.g., C0 < 0.5, meaning the normal data did not trigger a disaster judgment), then even if the enhanced criterion value E of the anomalous features is large, the corrected confidence level will not reach the judgment threshold, and no misjudgment will occur; for example, if C0 = 0.3 and E = 0.421, If C = 0.3 + 0.6 × 0.421 × 0.7 ≈ 0.477, it is still lower than the judgment threshold of 0.5.

[0041] Example 2: This embodiment, based on Embodiment 1, further illustrates the steps for adjusting abnormal weights and distinguishing abnormal causes.

[0042] Anomaly weight adjustment: The system maintains the historical reliability database of each monitoring station. In step S5 of Example 1, 10 abnormal nodes that meet the time-series correlation conditions are included in the enhancement criterion calculation (station B is not among these 10 nodes because it does not meet the time-series correlation conditions). The system calculates the reliability coefficient of the station according to formula (3). Its historical failure frequency and reliability coefficient are as follows (f max (Set to 12 times / year) Station A, f i =1 time / year, R i = 1 - 1 / 12 = 0.917; Station C, f i = 2 times / year, R i = 1 - 2 / 12 = 0.833; Station D, f i =8 times / year, R i = 1 - 8 / 12 = 0.333; Station F, f i =0 times / year, R i = 1 - 0 / 12 = 1.000; Station G, f i =3 times / year, R i = 1 - 3 / 12 = 0.750; Station H, f i=1 time / year, R i = 0.917; Station J, f i = 2 times / year, R i = 0.833; Station K, f i =0 times / year, R i = 1.000; Station L, f i =1 time / year, R i =0.917; Station M, f i =4 times / year, R i = 1 - 4 / 12 = 0.667; Among them, station D had the highest historical failure frequency (8 times / year), with a reliability coefficient of only 0.333, and its contribution of abnormal characteristics was significantly reduced.

[0043] according to Calculate the normalized value of the number of abnormal nodes, where N max =50: = 0.917 + 0.833 + 0.333 + 1.000 + 0.750 + 0.917 + 0.833 + 1.000 + 0.917 + 0.667 = 8.167 f(N a ) = 8.167 / 50 = 0.163 Compared to f(N) without weight adjustment in Implementation Example 1 a ) = 10 / 50 = 0.2, weighted f(N) a The value decreased from 0.2 to 0.163, reflecting the reduction effect of historically unstable stations (especially station D) on the enhancement criterion contribution.

[0044] g(S a ) and h(M t The values ​​are the same as in Example 1, namely 0.039 and 0.833. Substituting them into Equation (1), the weighting coefficients w1=0.4, w2=0.2, and w3=0.4. The weighted and adjusted enhanced criterion value E is calculated as follows: E = 0.4×0.163 + 0.2×0.039 + 0.4×0.833 = 0.065 + 0.008 + 0.333 =0.406 Compared to E=0.421 in Example 1 without weight adjustment, E is reduced to 0.406 after weight adjustment.

[0045] Substituting further into equation (2), C0 = 0.75, : C = 0.75 + 0.6×0.406×(1 - 0.75) = 0.75 + 0.061 = 0.811 The corrected confidence level is 0.811, which is slightly lower than 0.813 in Example 1 when no weight adjustment was performed. This reflects the suppressive effect of abnormal weight adjustment on the impact of historically unstable nodes. At the same time, the overall judgment result did not shift significantly, indicating that the weight adjustment mechanism maintains the system sensitivity while reducing the risk of misjudgment.

[0046] Distinguishing the causes of anomalies: After identifying the abnormal nodes, the system further obtains the communication link status information of the abnormal nodes. Among them, station E has data loss within the time window T, but its communication link status shows that the communication link was interrupted during this period and the interruption was caused by the failure of the upstream network equipment. At the same time, the environmental monitoring data (such as temperature, humidity, and power supply voltage) in the area where station E is located are all within the normal range. Based on this, the system determines that the data anomaly of station E is caused by a communication failure rather than a disaster event, and excludes it from the scope of the enhanced criteria generation.

[0047] Example 3: This embodiment illustrates the structure and function of the disaster event enhancement judgment system based on the abnormal characteristics of monitoring nodes provided by the present invention, such as... Figure 3 As shown, the system includes the following modules: Data acquisition module: Establishes communication connections with each monitoring node in the distributed monitoring network, receives event data and status data reported by each node in real time, performs time synchronization processing and preprocessing on the received data, and stores the processed data in the data cache area for subsequent modules to call.

[0048] Anomaly detection module: Reads the status data of each monitoring node from the data cache, and detects each node one by one according to the preset anomaly judgment rules. The anomaly judgment rules include signal interruption judgment rules, data anomaly judgment rules and mutation judgment rules. The threshold parameters of each rule can be configured by the user according to the specific application scenario. The anomaly detection module outputs a list of abnormal nodes, including the identifier, geographical location, anomaly type and anomaly occurrence time of each abnormal node.

[0049] The spatial analysis module receives the list of anomalous nodes output by the anomaly identification module, obtains the geographic coordinates of each anomalous node, and calculates the spatial distance between each anomalous node. When the number of anomalous nodes within a preset radius R is not less than a preset threshold N, it is determined that the anomalous nodes are concentrated in a specific area with the center of that area as the center and R as the radius. The spatial analysis module further calculates the spatial distance D between the center of this specific area and the center of the detected event. When D is less than a preset distance threshold D0, the spatial analysis module determines the spatial distance. maxIn this case, it is determined that the specific area has a spatial correlation with the area of ​​the detected event; or, it is determined whether the specific area is located within the preset buffer zone of the theoretical propagation path of the detected event, thereby determining the spatial correlation.

[0050] The time series analysis module receives a subset of spatially correlated abnormal nodes from the spatial analysis module, extracts the timestamp of each abnormal node, compares the time of abnormal occurrence with the initial time of the detected events, calculates the time difference, sorts each node according to its distance from the event center, analyzes the consistency between the time series of abnormal occurrences and the theoretical event propagation order, and outputs a time series matching index.

[0051] Criterion generation module: Combines the output results of the spatial analysis module and the temporal analysis module to determine whether the spatial distribution characteristics and temporal characteristics of the abnormal node simultaneously meet the preset conditions. When the conditions are met, the enhanced criterion value E is calculated according to the weighted fusion formula shown in equation (1). The weight coefficient and normalization function parameter of the criterion generation module can be configured by the user according to the specific application scenario.

[0052] Correction module: Receives the enhanced criterion value E output by the criterion generation module and the original judgment result from the existing disaster judgment system. The correction module calculates the corrected confidence level C according to the fusion formula shown in formula (2) and updates the risk level according to the correspondence between the confidence level and the risk level. The correction module outputs the corrected judgment result to the early warning release module or the superior management system.

[0053] The modules communicate with each other through standardized data interfaces. The overall system can be deployed on a centralized server platform or a distributed computing platform and supports real-time online operation.

[0054] 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 method for determining enhanced disaster events based on abnormal characteristics of monitoring nodes, characterized in that, Includes the following steps: Step S1, Data Acquisition and Basic Judgment: Acquire event data and status data from multiple monitoring nodes in the distributed monitoring network, filter and analyze the event data to obtain preliminary disaster judgment results; the event data includes physical quantity data collected by each monitoring node, and the status data includes the operating status information and data integrity information of each monitoring node; The preliminary disaster assessment results include the detected event area or event propagation path, the initial event time, and the original disaster assessment confidence level C0; Step S2, Abnormal Node Identification: Analyze the status data of the multiple monitoring nodes and identify abnormal nodes according to the preset abnormal judgment rules. The abnormal node is a node that meets at least one of the following conditions within a preset time window: no data output, output data exceeds the preset normal range, data changes or is interrupted. Step S3, Abnormal Spatial Distribution Analysis: Obtain the geographical location information of the abnormal nodes, analyze the spatial distribution characteristics of the abnormal nodes, determine whether the abnormal nodes are concentrated in a specific area, and determine the spatial correlation between the specific area and the detected event area or event propagation path. Step S4, Anomaly Timing Analysis: Extract the anomaly occurrence time of the anomaly node, analyze the temporal relationship between the anomaly occurrence time and the event initiation time, and the consistency between the anomaly occurrence time and the event propagation order; Step S5: Enhanced criterion generation: When the spatial distribution characteristics and temporal characteristics of the abnormal nodes meet the preset conditions, an enhanced criterion E is generated based on the number of abnormal nodes, the range of abnormal areas, and the time of abnormal occurrence. Step S6, Result Correction: Based on the obtained original disaster determination confidence level C0, under the condition that the original disaster determination confidence level C0 meets the preset initial determination threshold, the enhanced criterion E is introduced, and the original disaster determination result is corrected based on the enhanced criterion E to obtain the final disaster determination result; When the confidence level C0 of the original disaster determination is lower than the preset initial determination threshold, the enhancement criterion E is not used to correct the original disaster determination result.

2. The method according to claim 1, characterized in that, In step S2, the anomaly determination rules include: if the duration of continuous no data output by a monitoring node exceeds a preset threshold T1 within a preset time window T, the node is determined to be a signal interruption type anomaly node; if the output data value of a monitoring node exceeds a preset multiple standard deviation range of the historical data statistical distribution of the node within a preset time window T, the node is determined to be a data anomaly type anomaly node; if the rate of change of adjacent sampling points of a monitoring node exceeds a preset rate of change threshold within a preset time window T, the node is determined to be a sudden change type anomaly node.

3. The method according to claim 1, characterized in that, In step S3, the method for determining whether the abnormal nodes are spatially concentrated in a specific area includes: calculating the spatial distance between the abnormal nodes; when the number of abnormal nodes within a preset radius R is not less than a preset number threshold N, it is determined that the abnormal nodes are concentrated in a specific area with the center of the area as the center and R as the radius.

4. The method according to claim 1, characterized in that, In step S3, the method for determining the spatial correlation between the specific region and the detected event region or event propagation path includes: calculating the spatial distance D between the center of the specific region and the center of the detected event; when D is less than a preset distance threshold D... max When the specific region is spatially correlated with the region of the detected event, it is determined whether the specific region is located within a preset buffer zone of the theoretical propagation path of the detected event.

5. The method according to claim 1, characterized in that, In step S4, the analysis of the temporal relationship between the anomaly occurrence time and the event initiation time includes: extracting the timestamp t of the first occurrence of the anomaly at the anomaly node. a Obtain the initial start time t0 of the detected events. a The absolute value of the time difference between t0 and t0 is less than the preset time threshold. When the time of occurrence of the abnormal node is determined to be temporally correlated with the initial time of the event, the analysis of the consistency between the time of occurrence of the abnormal node and the order of event propagation includes: sorting each abnormal node from near to far according to the spatial distance between each abnormal node and the center of the detected event, and determining whether the timestamp of the first occurrence of the abnormal node monotonically increases or approximately monotonically increases with the increase of distance. When the proportion of nodes that meet the monotonically increasing condition exceeds the preset matching threshold, it is determined that the time of occurrence of the abnormal node and the order of event propagation are consistent.

6. The method according to claim 1, characterized in that, In step S5, the method for generating the enhanced criterion includes: based on the number of abnormal nodes N a Abnormal area range S a Matching degree M with abnormal time series t The enhancement criterion value E is calculated according to the preset weighted fusion formula: E = w1×f(N a ) + w2×g(S a ) + w3×h(M t ) (1) Where w1, w2, and w3 are preset weight coefficients, and w1 + w2 + w3 = 1; f(N a ) is the normalization function for the number of abnormal nodes; g(S) a h(M) is the normalization function for the range of the abnormal region; t ) is the normalization function for the abnormal time series matching degree.

7. The method according to claim 1, characterized in that, In step S6, the method for correcting the confidence level of the disaster event includes: fusing the original disaster determination confidence level C0 with the enhanced criterion value E to obtain the corrected confidence level C. （2） in, The preset enhancement coefficient, It is used to control the degree to which the enhancement criterion contributes to the final judgment result.

8. The method according to claim 6, characterized in that, In step S5, before calculating the enhancement criterion value E, an anomaly weight adjustment step is also included: weighting and reducing the contribution of each anomaly node to its anomaly characteristics based on the historical failure frequency of each node; specifically, calculating the reliability coefficient R of each anomaly node. i : R i = 1 - f i / f max (3) Among them, f i f is the historical failure frequency of the i-th node. max For nodes with a high historical failure frequency, the reliability coefficient R is set to the preset maximum failure frequency threshold. i The smaller the value, the lower the weight of anomalous features; for nodes with a history of stable operation, the reliability coefficient R... i The larger the value, the higher the weight of the abnormal features; The normalization function for the number of abnormal nodes in formula (1) is f(N). a When ), the reliability coefficient R of each abnormal node is used. i We replace direct counting with weighted summation, i.e. The summation range is the range of abnormal nodes that satisfy the temporal correlation condition.

9. The method according to claim 1, characterized in that, Between steps S2 and S3, there is also an abnormal cause differentiation step: obtaining the communication link status information and environmental monitoring data of the abnormal node. When the communication link of the abnormal node is interrupted during the abnormal occurrence period and the environmental monitoring data of the area where the node is located is within the normal range, it is determined that the data abnormality of the abnormal node is caused by a communication failure, and the abnormal node is excluded from the abnormal nodes identified in step S2 so that it does not participate in the analysis of step S3 and subsequent steps.

10. A disaster event enhancement judgment system based on abnormal characteristics of monitoring nodes, characterized in that, include: The data acquisition module is used to acquire event data and status data from multiple monitoring nodes in the distributed monitoring network. An anomaly identification module is used to analyze the status data of the multiple monitoring nodes according to preset anomaly judgment rules and identify abnormal nodes. The spatial analysis module is used to obtain the geographical location information of the abnormal nodes, analyze the spatial distribution characteristics of the abnormal nodes, and determine the spatial correlation between the spatial distribution of the abnormal nodes and the detected event area or event propagation path. The timing analysis module is used to extract the occurrence time of the abnormal node, analyze the timing relationship between the occurrence time of the abnormal node and the initial time of the event, and the consistency with the event propagation order. The criterion generation module is used to generate an enhanced criterion E based on the number of abnormal nodes, the range of abnormal regions, and the time of occurrence of abnormalities when the spatial distribution characteristics and temporal characteristics of the abnormal nodes simultaneously meet preset conditions. The correction module is used to, based on the obtained original disaster determination confidence level C0, and under the condition that the original disaster determination confidence level C0 meets the preset initial determination threshold, introduce the enhancement criterion E, and correct the original disaster determination result based on the enhancement criterion E to obtain the final disaster determination result; When the confidence level C0 of the original disaster determination is lower than the preset initial determination threshold, the enhancement criterion E is not used to correct the original disaster determination result.

Images