Mountain fire disaster risk assessment method based on interaction relationship between mountain fire and electric power facilities

Abstract

The invention discloses a mountain fire disaster risk assessment method based on the interaction relationship between mountain fire and electric power facilities, and particularly relates to the field of electric power system safety monitoring and disaster risk assessment, which comprises the following steps: acquiring monitoring data by using satellite remote sensing equipment, an unmanned aerial vehicle observation device and a ground meteorological sensor; performing space-time correction on the monitoring data of each source according to the timestamps and the geographic coordinates, adjusting the sampling density according to the related weights, and outputting a monitoring data sequence; according to the invention, through constructing a multi-level data closed-loop process integrating multi-source monitoring, dynamic transmission, health prediction and path reconstruction, continuous acquisition of monitoring data, adaptive evaluation of node states and real-time coupling calculation of risk results are realized; the problem of risk assessment inaccuracy caused by unstable data transmission and fault prediction lag in the prior art is solved.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of power system safety monitoring and disaster risk assessment, more specifically, the present application relates to a mountain fire disaster risk assessment method based on the interaction relationship between mountain fire and power facilities. BACKGROUND

[0002] In the current power grid mountain fire monitoring and disaster risk assessment system, the industry generally adopts a hierarchical structure of multi-source remote sensing collection, edge computing early warning and cloud fusion evaluation to ensure the safe operation of power facilities in the mountain fire environment. However, the existing technology still has fundamental defects in the integration of data transmission and fault prediction health management chain. First, due to the instability of the data transmission chain, under the action of high temperature radiation, smoke diffusion and terrain shielding, the signal attenuation between satellite link and ground microwave station is significant, resulting in sudden delay and breakage of spatio-temporal data flow, especially forming data voids during the most intense period of the fire. The traditional redundant channel switching mechanism is subject to passive triggering and fixed priority, and cannot realize real-time reconstruction under disaster conditions, directly weakening the continuity and credibility of the data flow. Second, due to the chain erosion of data anomalies to fault prediction health management, when the transmission layer appears time tearing and data drift, the system model for predicting power line arc, transformer local overheating, equipment insulation deterioration and other hidden dangers will receive mismatched signals, causing lagging prediction errors. The model self-checking module misjudges the system state as healthy because the perceived input features are still in the statistical stationary interval, thereby inhibiting the triggering of self-correction and retraining mechanism. Finally, due to the failure propagation of system-level health management, once the data transmission error is continuously amplified, the cloud risk assessment module will update the weights based on the distorted input, and the edge side self-diagnosis system will lose the reference benchmark, causing prediction deviation accumulation, alarm threshold drift and false stability of health score. Therefore, in the existing power grid mountain fire risk assessment system, the discontinuity and quality degradation of data transmission directly destroy the real-time closed loop of fault prediction and health management, making the system lose dynamic perception and self-healing ability in the most critical disaster stage, thereby failing to maintain the safe operation of power facilities in the mountain fire environment. SUMMARY

[0003] In order to overcome the above-mentioned defects of the prior art, the embodiments of the present application provide a mountain fire disaster risk assessment method based on the interaction relationship between mountain fire and power facilities, which realizes continuous collection of monitoring data, adaptive assessment of node state and real-time coupled calculation of risk results by constructing a multi-level data closed loop process integrating multi-source monitoring, dynamic transmission, health prediction and path reconstruction, to solve the risk assessment misalignment problem caused by unstable data transmission and lagging fault prediction in the prior art.

[0004] To achieve the above object, the present application provides the following technical scheme: a mountain fire disaster risk assessment method based on the interaction relationship between mountain fire and power facilities, comprising: S1, by using satellite remote sensing equipment, unmanned aerial vehicle observation device and ground meteorological sensor to obtain monitoring data, the monitoring data from different sources is corrected in space and time according to time stamp and geographic coordinates, and the sampling density is adjusted according to the related weight, and the monitoring data sequence is output; S2, by inputting the monitoring data sequence into the multi-channel data transmission chain to execute dynamic transmission, the node signal strength, time delay and packet loss rate are counted, the deviation of instantaneous stability index and historical stability threshold is compared, the channel switching condition is updated and the stable transmission sequence is output; S3, by inputting the stable transmission sequence into the fault prediction and health management process, the degradation rate value is calculated by using signal variance and attenuation slope, the autoregressive prediction is executed to update the node health score, the continuous decline interval is calibrated and the node health record table is output; S4, by inputting the node health record table into the path reconstruction process, the replacement direction is calculated according to the signal energy gradient of adjacent nodes, the node transmission weight distribution is adjusted, the interpolation compensation and topology update are executed for the mismatch interval, and the repair data sequence is output; S5, by inputting the repair data sequence into the power facility mountain fire risk assessment process, the feature matrix is constructed according to the node health aggregation vector and its normalized node health aggregation weight calculated according to the path level index and the monitoring data, the coupling result of mountain fire heat source intensity, power load and environmental humidity difference value is calculated, and the comprehensive risk assessment result is output.

[0005] In a preferred embodiment, in S1, the ground surface temperature distribution and fire heat radiation intensity data are obtained by using satellite remote sensing equipment, the power facility appearance thermal imaging and smoke plume diffusion image data are obtained by using unmanned aerial vehicle observation device, and the ground surface temperature, humidity and wind speed data are obtained by using ground meteorological sensor, the data are collected at the same time to form an initial monitoring data set; The initial monitoring data set is sorted according to the collection time, the average time difference of adjacent sampling intervals is calculated and linear interpolation is performed to correct the difference of sampling period of different equipment, and the space alignment is performed according to the geographic coordinates to generate the monitoring data matrix under the unified time and space reference; By calculating the correlation coefficient of ground surface temperature, humidity and wind speed in the monitoring data matrix, the correlation coefficient is used as a sampling adjustment factor to perform weighted adjustment on the sampling point spacing, so that the sampling points in high correlation area are kept dense and the sampling points in low correlation area are moderately sparse, and the space-time balanced monitoring data distribution is obtained; The continuous sampling point sequence is extracted from the monitoring data distribution in time index order, the monitoring data of ground surface temperature, humidity and wind speed are respectively extracted into continuous time series according to the collection time index, the change rate of each physical quantity is calculated by dividing the change of the observation value of the adjacent sampling points in the time series by the sampling time interval, the change rate result is fitted into a continuous function according to the time index, the corresponding time trend curve is obtained, the time trend curve is taken as the interpolation reference to perform data completion on the missing sampling points, and the continuity of the data is tested through residual constraint, and the monitoring data sequence after time correction, spatial registration and density optimization is output.

[0006] In a preferred embodiment, in S2, the monitoring data sequence is further input into a multi-channel data transmission chain, parallel transmission paths are established between satellite links, ground optical fiber links and wireless relay nodes; the signal strength, transmission delay and packet loss number of each node are collected in each transmission period to form a node operation state table; The node operation state table is recursively calculated according to a fixed time window, and in each time window, the signal strength change rate is calculated by dividing the difference between the current time signal strength and the previous window signal strength by the time interval of the two windows; the delay drift rate is calculated by dividing the difference between the current time delay mean value and the previous window delay mean value by the time interval of the two windows; The signal strength change rate and the delay drift rate of all nodes are normalized to generate an instantaneous stability index, and a deviation comparison is performed with the historical stability threshold of the corresponding node: When the node signal strength is less than the preset signal threshold and the delay drift rate is greater than the preset drift threshold, the node is marked as an unstable node; If the above conditions are not met, the node weight remains unchanged and the node operation state table is returned to continue monitoring.

[0007] In a preferred embodiment, in S2, when the node is marked as an unstable node, the weighted average of the signal strength decay rate, the packet loss rate and the delay drift rate is calculated to generate a node risk score; When the node risk score exceeds the channel switching threshold and the packet loss rate is higher than the preset packet loss threshold, the path rearrangement judgment process is entered; if any condition is not met, the instantaneous stability index calculation process is re-executed for recursive calculation; In the path rearrangement judgment process, a neighborhood candidate node set is constructed based on the node operation state table, the neighborhood candidate node set includes adjacent nodes that have direct communication links with the unstable node; the signal strength, delay mean value and packet loss rate of the neighborhood candidate nodes are comprehensively calculated, and the node transmission weight is generated according to the weighted results of the node signal strength and the delay reciprocal, and the packet loss rate is inversely proportional; the node stability score is calculated by the node transmission weight and the signal strength together. When the stability score of any one of the neighborhood candidate nodes is greater than the risk score of the unstable node, the node corresponding to the upper limit of the stability score is selected as the replacement node, channel switching is performed, and the transmission path structure is updated; if the condition is not met, the original channel configuration is maintained, and the unstable node is marked as a continuously monitored node; In the three consecutive transmission periods after channel switching, when the node signal strength is greater than the preset signal threshold and the time delay drift rate is less than the preset drift threshold, the unstable label is removed, and the node running state table is returned to execute the next period of statistics; If any condition is not met, the node risk score calculation process continues to be executed until the node state is restored or replaced, and finally the dynamically adjusted stable transmission sequence is output.

[0008] In a preferred embodiment, in S3, the stable transmission sequence is also segmented according to the node and time sequence to form the signal strength sequence, the transmission time delay sequence, and the packet loss rate sequence of each node; The signal strength sequence is extracted by a fixed time window to obtain a sampling point, the signal strength of adjacent sampling points is differentiated, and the sampling time interval is divided to obtain the signal strength change rate; The average value of the signal strength change rate sequence is calculated, and the square of the difference between each signal strength change rate and the average value is averaged to obtain the signal change variance; The least squares fitting is performed with the time index as the independent variable and the signal strength as the dependent variable, and the slope of the fitted straight line is taken as the signal attenuation slope; The node degradation rate value is generated by normalizing the signal change variance and the signal attenuation slope; When the node degradation rate value is greater than the preset degradation threshold, the node is determined to be a performance degradation node; When the node degradation rate value is less than or equal to the preset degradation threshold, the node state is maintained, and the signal strength sequence, the transmission time delay sequence, and the packet loss rate sequence of the node are returned to continue the differentiation and fitting calculation.

[0009] In a preferred embodiment, in S3, the signal change variance, the signal attenuation slope, and the time delay drift rate of the performance degradation node are arranged according to the time index to form a prediction input sequence, the autoregressive calculation is performed on the prediction input sequence, the node health score prediction value of the future continuous transmission period is solved, and the health score offset is calculated by the difference between the current node health score and the node health score prediction value; When the health score offset remains negative growth in the continuous three prediction periods and the absolute value exceeds the preset health decline threshold, the node is marked as a continuous decline interval; if the above conditions are not met at the same time, the node degradation rate value is recalculated; As an alternative determination condition, when the health score offset is negative growth in two consecutive prediction periods and the average decline amplitude is greater than the preset health decline threshold, it is determined that the node is in a continuous decline interval; if any condition is not met, return to perform autoregressive prediction operation to update the health score prediction value; When the node is determined to be in a continuous decline interval, the node health score, node degradation rate value, signal change variance, and the corresponding relationship of the continuous decline interval are recorded as a node health record table, and the node health record table is output.

[0010] In a preferred embodiment, in S4, further comprising reconstructing the path by inputting the node health record table according to the node number and the time index input path, extracting the performance degradation nodes that have been calibrated and their corresponding signal strength, transmission delay and packet loss rate data from the node health record table; Determine the node with a health score higher than the preset health threshold and a signal strength greater than the signal threshold as a candidate healthy node, and form a performance degradation node set and a candidate healthy node set respectively; For each performance degradation node in the performance degradation node set, calculate the signal strength difference between the candidate healthy nodes with direct communication link to it divided by the distance between nodes to obtain the adjacent node signal energy gradient; When the adjacent node signal energy gradient is greater than the preset energy gradient threshold, determine that the direction is the alternative direction; if the adjacent node signal energy gradient is less than or equal to the energy gradient threshold, keep the original path connection and return to the node health record table to recalculate the signal energy gradient.

[0011] In a preferred embodiment, in S4, further comprising, after determining the alternative direction, performing node transmission capacity calculation according to the signal strength, transmission delay and packet loss rate of the candidate healthy node, defining the node transmission capacity as a weighted function of signal strength and delay inverse, packet loss rate inverse; Perform node transmission weight adjustment according to the joint results of node transmission capacity and signal energy gradient, decrease the transmission weight of performance degradation nodes by energy gradient proportion, increase the transmission weight of candidate healthy nodes by the same proportion, and generate an updated node transmission weight distribution table; Perform interpolation compensation on the time section with data gaps in the node transmission weight distribution table, use the time index and signal strength change rate of the candidate healthy node to obtain the interpolation point value, and perform path topology structure update based on the interpolation result; The average signal strength is obtained by calculating the difference between the average signal strength of the updated path in the same time interval and the average signal strength before updating, and when the average signal strength of the updated path is greater than the preset topology stability threshold, the repair data sequence is output; if the average signal strength is less than or equal to the topology stability threshold, the node transmission weight adjustment step is returned to continue to execute the alternative direction optimization.

[0012] In a preferred embodiment, in S5, the repair data sequence is also input into the power facility wildfire risk assessment process, the health score, signal strength and transmission delay data of each node are extracted, and hierarchical weighting calculation is performed according to the path level index of the node in the transmission chain topology; The nodes in the transmission chain are sorted according to the path level index, and the hierarchical index value is used as the node hierarchical weight factor; when the hierarchical index value is less than a preset hierarchical threshold, it is defined as an upper layer node, and when the hierarchical index value is greater than or equal to the threshold, it is defined as a lower layer node; The weighted coefficient is determined according to the reciprocal function of the hierarchical weight factor, the weighted accumulation of the health score of each node is performed to generate a health aggregation vector, and the health aggregation vector is normalized to obtain a node health aggregation weight; The health aggregation vector and the ground temperature, fire source heat radiation intensity, power facility load and environmental humidity data in the monitoring data sequence are aligned according to the time index, spatial registration is performed according to the geographical coordinates corresponding to the nodes, and a feature matrix composed of the node health aggregation vector component and the environmental monitoring data difference of the ground temperature, fire source heat radiation intensity, power facility load and environmental humidity is constructed; In the feature matrix, the node health aggregation weight is used as the environmental monitoring data correction coefficient, and the weighted correction of the influence proportion of the fire source heat radiation intensity, power load and environmental humidity difference is performed; The corrected environmental monitoring data is used as the input to perform coupling calculation between the fire source heat radiation intensity, power load and environmental humidity difference to obtain a coupling relationship index.

[0013] In a preferred embodiment, in S5, when the coupling relationship index is greater than a preset risk threshold, the time interval is marked as a high risk interval; if the coupling relationship index is less than or equal to the risk threshold, it is marked as a low risk interval; The proportion of the number of high risk intervals to the total number of evaluation intervals in the continuous time window is counted, and the mean difference of the node health aggregation vector in adjacent time windows is divided by the time interval to obtain the node health score change rate; The high risk interval proportion and the node health score change rate are linearly combined according to a set weight coefficient to generate a comprehensive risk value, and a comprehensive risk assessment result is output; When the comprehensive risk value keeps rising in three consecutive time windows and the high-risk interval ratio exceeds the set threshold, the power facility health warning mechanism is triggered. If the condition is not met, the feature matrix updating process is re-executed to correct the evaluation result.

[0014] Technical effects and advantages of the present application: 1. The present application introduces dynamic stability calculation and adaptive channel switching mechanism in the multi-channel data transmission chain, which can monitor signal strength, transmission delay and packet loss rate in real time under extreme environments such as high temperature radiation, smoke diffusion and terrain shielding, and perform active reconstruction according to the deviation of instantaneous stability and historical stability threshold, which fundamentally solves the data interruption problem caused by passive switching of the transmission chain in the prior art, and realizes the continuity and reliability of power facility monitoring data flow under disaster conditions. 2. By establishing a node operating state table and introducing a joint recursive algorithm of signal strength change rate and time delay drift rate, the present application can identify unstable nodes in advance during transmission, and realize real-time replacement of neighbor nodes according to risk score, so as to realize self-repair at the initial stage of link damage or signal attenuation and improve the robustness of power communication link. 3. The present application calculates the node degradation rate by signal variance and attenuation slope, and generates health score prediction value by using autoregressive model, which realizes the early determination of node performance degradation trend and avoids the accumulation of hidden faults caused by lag detection in traditional methods. 4. By establishing a joint scheduling mechanism based on signal energy gradient and node transmission weight, the present application can realize adaptive migration of transmission load between performance degradation nodes and healthy nodes, and repair data gaps by combining interpolation compensation algorithm, which ensures the structural stability of the overall network topology after disturbance. 5. The present application realizes the dynamic coupling of power equipment health status and mountain fire environment parameters by weighted aggregation of node health score at transmission chain level, constructing a feature matrix containing fire source heat radiation intensity, power load and environmental humidity difference, and correcting the influence proportion of environmental variables by health aggregation weight, which improves the reliability of comprehensive risk assessment. BRIEF DESCRIPTION OF DRAWINGS

[0015] Figure 1 The method step flowchart of the present application. DETAILED DESCRIPTION

[0016] The technical solutions in the embodiments of the present application will be described clearly and completely in combination with the drawings in the embodiments of the present application. Obviously, the described embodiments are only part of the embodiments of the present application, not all. Based on the embodiments in the present application, all other embodiments obtained by those skilled in the art without creative labor are within the scope of protection of the present application.

[0017] The accompanying drawings are referred to in Figure 1 The application provides a wildfire disaster risk assessment method based on interaction between wildfire and power facilities. S1, obtaining monitoring data by using satellite remote sensing equipment, unmanned aerial vehicle observation devices and ground meteorological sensors, performing space-time correction on the monitoring data from different sources according to time stamps and geographic coordinates, adjusting the sampling density according to the related weight, and outputting a monitoring data sequence; S2, inputting the monitoring data sequence into a multi-channel data transmission chain to perform dynamic transmission, counting node signal strength, time delay and packet loss rate, comparing the deviation of the instantaneous stability index from the historical stability threshold, updating the channel switching condition and outputting a stable transmission sequence; S3, inputting the stable transmission sequence into a fault prediction and health management process, calculating the degradation rate value by using signal variance and attenuation slope, updating the node health score by performing autoregressive prediction, calibrating the continuous decline interval and outputting a node health record table; S4, inputting the node health record table into a path reconstruction process, calculating a replacement direction according to the signal energy gradient of adjacent nodes, adjusting the node transmission weight distribution, performing interpolation compensation and topology update on the mismatch interval, and outputting a repaired data sequence; S5, inputting the repaired data sequence into a power facility wildfire risk assessment process, constructing a feature matrix according to the node health aggregation vector and the normalized node health aggregation weight obtained by calculating the path level index and the monitoring data, calculating the coupling result of the wildfire heat source intensity, the power load and the environmental humidity difference, and outputting the comprehensive risk assessment result.

[0018] In S1, the satellite remote sensing equipment is used to obtain the ground temperature distribution and the heat radiation intensity data of the fire source, the unmanned aerial vehicle observation device is used to obtain the power facility appearance thermal imaging and smoke plume diffusion image data, and the ground meteorological sensor is used to obtain the ground temperature, humidity and wind speed data. The data are collected simultaneously to form an initial monitoring data set; The initial monitoring data set is sorted according to the collection time, the average time difference between adjacent sampling intervals is calculated and linear interpolation is performed to correct the sampling period difference of different devices, and the space alignment is performed according to the geographic coordinates to generate a monitoring data matrix under the unified time and space reference; The correlation coefficients of the ground temperature, humidity and wind speed in the monitoring data matrix are calculated, the correlation coefficients are used as sampling adjustment factors, the sampling point spacing is adjusted, the sampling points in the high correlation area are kept dense, and the sampling points in the low correlation area are moderately sparse, so that the monitoring data distribution is balanced in time and space; The continuous sampling point sequence is extracted from the monitoring data distribution in time index order, the monitoring data of ground temperature, humidity and wind speed are respectively extracted into continuous time series according to the collection time index, the change rate of each physical quantity is calculated by dividing the change of the observation value of the adjacent sampling points in the time series by the sampling time interval, the change rate result is fitted into a continuous function according to the time index, the corresponding time trend curve is obtained, the time trend curve is taken as the interpolation reference to perform data completion on the missing sampling points, and the continuity of the data is tested through residual constraint, and the monitoring data sequence after time correction, space registration and density optimization is output.

[0019] In S2, the monitoring data sequence is also input into a multi-channel data transmission chain, and parallel transmission paths are established between satellite links, ground optical fiber links and wireless relay nodes; the signal strength, transmission delay and packet loss number of each node are collected in each transmission period to form a node operation state table; The node operation state table is recursively calculated according to a fixed time window, and in each time window, the signal strength change rate is calculated by dividing the difference between the current time signal strength and the previous window signal strength by the time interval of the two windows; the delay drift rate is calculated by dividing the difference between the current time delay mean value and the previous window delay mean value by the time interval of the two windows; The signal strength change rate and delay drift rate of all nodes are normalized to generate an instantaneous stability index, and the deviation is compared with the historical stability threshold of the corresponding node: When the node signal strength is less than the preset signal threshold and the delay drift rate is greater than the preset drift threshold, the node is marked as an unstable node; If the above conditions are not met, the node weight remains unchanged and the node operation state table is returned to continue monitoring.

[0020] In S2, when the node is marked as an unstable node, the weighted average of the signal strength decay rate, packet loss rate and delay drift rate is calculated to generate a node risk score; When the node risk score exceeds the channel switching threshold and the packet loss rate is higher than the preset packet loss threshold, the path rearrangement judgment process is entered; if any condition is not met, the instantaneous stability index calculation process is executed recursively; In the path rearrangement judgment process, a neighborhood candidate node set is constructed based on the node operation state table, and the neighborhood candidate node set includes adjacent nodes that have direct communication links with the unstable node; the signal strength, delay mean value and packet loss rate of the neighborhood candidate nodes are comprehensively calculated, and the node transmission weight is generated according to the weighted results of the node signal strength and the reciprocal of the delay, and the packet loss rate is inversely proportional; the node stability score is calculated by the node transmission weight and the signal strength together. When the stability score of any one of the neighborhood candidate nodes is greater than the risk score of the unstable node, the node corresponding to the upper limit of the stability score is selected as the replacement node, channel switching is performed, and the transmission path structure is updated; if the condition is not met, the original channel configuration is maintained, and the unstable node is marked as a continuously monitored node; In the three consecutive transmission periods after channel switching, when the node signal strength is greater than the preset signal threshold and the time delay drift rate is less than the preset drift threshold, the unstable label is removed, and the node running state table is returned to execute the next period of statistics; If any condition is not met, the node risk score calculation process continues to be executed until the node state is restored or replaced, and finally the dynamically adjusted stable transmission sequence is output.

[0021] In S3, the stable transmission sequence is also segmented according to the node and time sequence to form the signal strength sequence, the transmission time delay sequence, and the packet loss rate sequence of each node; The signal strength sequence is extracted by a fixed time window to obtain a sampling point, the signal strength of adjacent sampling points is differentiated, and the sampling time interval is divided to obtain the signal strength change rate; The average value of the signal strength change rate sequence is calculated, and the square of the difference between each signal strength change rate and the average value is averaged to obtain the signal change variance; The least squares fitting is performed with the time index as the independent variable and the signal strength as the dependent variable, and the slope of the fitted straight line is taken as the signal attenuation slope; The node degradation rate value is generated by normalizing the signal change variance and the signal attenuation slope; When the node degradation rate value is greater than the preset degradation threshold, the node is determined to be a performance degradation node; When the node degradation rate value is less than or equal to the preset degradation threshold, the node state is maintained, and the signal strength sequence, the transmission time delay sequence, and the packet loss rate sequence of the node are returned to continue the differentiation and fitting calculation.

[0022] In S3, the signal change variance, the signal attenuation slope, and the time delay drift rate of the performance degradation node are also arranged according to the time index to form a prediction input sequence, the autoregressive calculation is performed on the prediction input sequence, the node health score prediction value of the future continuous transmission period is solved, and the health score offset is calculated by the difference between the current node health score and the node health score prediction value; When the health score offset maintains negative growth in the continuous three prediction periods and the absolute value exceeds the preset health decline threshold, the node is marked as a continuous decline interval; if the above conditions are not met at the same time, the node degradation rate value is recalculated; As an alternative judgment condition, when the health score offset is negative for two consecutive prediction periods and its average decrease is greater than the preset health decline threshold, the node is determined to be in the continuous decline interval; if any condition is not met, the autoregressive prediction calculation is re-executed to update the predicted health score value. Once a node is identified as being in a continuously declining interval, the node health score, node degradation rate value, signal variation variance, and the correspondence with the continuously declining interval are recorded in a node health record table. The node health record table is then output for subsequent path reconstruction and transmission optimization.

[0023] S4 also includes a process of reconstructing the node health record table by inputting the node number and time index into the path, and extracting the identified performance degradation nodes and their corresponding signal strength, transmission delay and packet loss rate data from the node health record table; Nodes with health scores higher than a preset health threshold and signal strength greater than a signal threshold are identified as candidate healthy nodes, forming a set of performance-degraded nodes and a set of candidate healthy nodes, respectively. For each degraded node in the set of degraded nodes, calculate the signal strength difference between it and the candidate healthy nodes with which it has a direct communication link, divide it by the distance between the nodes, and obtain the signal energy gradient of the adjacent nodes. When the signal energy gradient of an adjacent node is greater than the preset energy gradient threshold, the direction is determined as the alternative direction; if the signal energy gradient of an adjacent node is less than or equal to the energy gradient threshold, the original path connection is maintained and the node health record table is returned to recalculate the signal energy gradient.

[0024] S4 also includes, after determining the alternative direction, performing node transmission capacity calculation based on the signal strength, transmission delay and packet loss rate of the candidate healthy nodes, defining the node transmission capacity as a weighted function that is inversely proportional to the signal strength and the reciprocal of the delay and the packet loss rate; The node transmission weight is adjusted based on the joint result of node transmission capacity and signal energy gradient. The transmission weight of the degraded node is reduced proportionally to the energy gradient, and the transmission weight of the candidate healthy node is increased proportionally to the same ratio, generating an updated node transmission weight allocation table. For time segments with data gaps in the node transmission weight allocation table, interpolation compensation is performed. The interpolation point value is obtained by using the time index of the candidate healthy node and the signal strength change rate, and the path topology is updated based on the interpolation result. The average signal strength is obtained by calculating the difference between the average signal strength of the updated path and the average signal strength before the update within the same time interval. When the improvement in the average signal strength of the updated path is greater than the preset topology stability threshold, the repair data sequence is output; if the improvement in the average signal strength is less than or equal to the topology stability threshold, the node transmission weight adjustment step is returned to continue the alternative direction optimization.

[0025] In S5, the repaired data sequence is also input into the power facility wildfire risk assessment process, the health scores, signal strengths, and transmission delay data of each node are extracted, and hierarchical weighting calculation is performed according to the path level index of the node in the transmission chain topology; The nodes in the transmission chain are sorted according to the path level index, and the hierarchical index value is used as the node hierarchical weight factor; when the hierarchical index value is less than a preset hierarchical threshold, it is defined as an upper layer node, and when the hierarchical index value is greater than or equal to the threshold, it is defined as a lower layer node; The weighted coefficient is determined according to the reciprocal function of the hierarchical weight factor, the weighted accumulation of the health scores of each node is performed, the health aggregation vector representing the overall operation state of the transmission chain is generated, and the health aggregation vector is normalized to obtain the node health aggregation weight; The health aggregation vector and the ground temperature, fire source heat radiation intensity, power facility load, and environmental humidity data in the monitoring data sequence are aligned according to the time index, spatial registration is performed according to the geographical coordinates corresponding to the nodes, and a feature matrix composed of the node health aggregation vector components and the environmental monitoring data of the ground temperature, fire source heat radiation intensity, power facility load, and environmental humidity difference is constructed; In the feature matrix, the node health aggregation weight is used as the environmental monitoring data correction coefficient, and the weighted correction of the influence proportion of the fire source heat radiation intensity, power load, and environmental humidity difference is performed; The corrected environmental monitoring data is used as input to perform coupling calculation between the fire source heat radiation intensity, power load, and environmental humidity difference, and the coupling relationship index is obtained.

[0026] In S5, when the coupling relationship index is greater than a preset risk threshold, the time interval is marked as a high-risk interval; if the coupling relationship index is less than or equal to the risk threshold, it is marked as a low-risk interval; The proportion of the number of high-risk intervals to the total number of evaluation intervals is counted in the continuous time window, and the mean difference of the node health aggregation vector in adjacent time windows is divided by the time interval to obtain the node health score change rate; The high-risk interval proportion and the node health score change rate are linearly combined according to the set weight coefficient to generate a comprehensive risk value, and the comprehensive risk assessment result is output; When the comprehensive risk value maintains an upward trend in three consecutive time windows and the high-risk interval proportion exceeds the set threshold, the power facility health warning mechanism is triggered; If the conditions are not met, the feature matrix updating process is re-executed to correct the evaluation result.

[0027] In specific implementation, the scheme first synchronously collects temperature, humidity, wind speed, fire heat radiation and other multi-source monitoring data through satellite remote sensing, unmanned aerial vehicle and ground meteorological sensor, constructs a basic monitoring data matrix, which is converted into a monitoring data sequence after time and space double correction, to ensure that each data has consistent time and space reference; then, the data is sent to a multi-channel data transmission chain, parallel paths are formed between satellite link, ground optical fiber and wireless relay, and transmission performance is described in real time with three indexes of signal strength, transmission delay and packet loss rate, the change trend is calculated through recursive calculation of fixed time window, the instantaneous stability index is generated, and the transmission chain is judged whether there is an unstable node by comparing with the historical threshold value; After the node transmission anomaly is confirmed, the system further calculates the weighted average value of signal strength attenuation rate, time delay drift rate and packet loss rate, generates node risk score, and judges whether to trigger path rearrangement according to the comparison of risk score and channel switching threshold value. If the risk is high, the system constructs a neighbor candidate node set, calculates the node transmission weight according to the signal strength, average time delay and packet loss rate of adjacent nodes, and then calculates the node stability score together with the signal strength, so as to select the replaceable healthy node and dynamically update the path structure; When the transmission is stable, the fault prediction and health management stage is entered, at this time, the signal strength sequence, transmission time delay sequence and packet loss rate sequence of each node are analyzed segmentally, the signal strength change rate and signal change variance are calculated by comparing the difference between sampling points, and the signal attenuation slope is fitted by least square method. These results together constitute the node degradation rate value, which is used to identify the performance degradation node; then the node health score in future periods is predicted through autoregressive analysis, and the health score offset is calculated; when the health score continuously decreases in multiple periods and exceeds the health decline threshold value, the node is marked as continuous decline interval, and the corresponding result is written into the node health record table to provide decision basis for path reconstruction; Next, the system reads the node health record table in the path reconstruction stage, extracts the performance degradation nodes and candidate healthy nodes that meet the conditions, calculates the signal energy gradient by comparing the signal strength difference and the distance between nodes; when the gradient exceeds the energy gradient threshold value, the replacement direction is determined; then, the system calculates the node transmission capacity according to the signal strength, transmission time delay and packet loss rate of the candidate node, and takes it as the weight adjustment basis to reduce the transmission weight of the degradation node and improve the transmission weight of the healthy node, forming a new node transmission weight distribution table; in the section with data gap, the continuity is restored through interpolation compensation, and the path topology structure is updated; when the improvement amount of average signal strength of the path exceeds the topology stability threshold value, the repaired data sequence is output, otherwise it returns to re-adjustment; Finally, in the risk assessment stage, the repaired data is input into the power facility wildfire risk assessment process. The system performs hierarchical weighting calculation based on the path level index of the node in the transmission chain topology, generates the node health aggregation vector, and obtains the node health aggregation weight through normalization. These weights, together with environmental monitoring data such as ground temperature, fire source heat radiation intensity, power load, and environmental humidity, form a feature matrix. In the feature matrix, the node health aggregation weight is used as a correction coefficient for the environmental monitoring data. The corrected environmental data is then used in the coupling calculation to obtain the coupling relationship index. If the index exceeds the risk threshold, it is marked as a high-risk interval, otherwise it is a low-risk interval. The system further determines the risk trend based on the comprehensive value of the high-risk interval proportion and the node health score change rate. When the risk rises in three consecutive time windows and the high-risk proportion exceeds the threshold, the system automatically triggers the power facility health warning mechanism. In actual application scenarios, the scheme makes the power grid wildfire risk assessment system shift from passive response to active prediction, and can achieve stable operation without human intervention in disaster-prone areas. The scheme uses multi-channel transmission chains and automatic path reconstruction mechanisms to automatically switch communication paths when satellite links are disturbed or ground nodes are damaged, ensuring continuous data flow and solving the problem of real-time monitoring interruption caused by unstable data transmission in wildfire sites. Moreover, the scheme uses dual parameters of degradation rate and health score offset to realize real-time perception of device status, solving the problem of inaccurate device health prediction. Finally, the scheme introduces node health aggregation weight correction of environmental impact proportion to make the assessment results more consistent with actual operation risks, solving the problem of disconnection between risk assessment and site status.

[0028] Finally: The above only describes the preferred embodiments of the present application and is not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principles of the present application shall be included in the protection scope of the present application.

Claims

1. A method for assessing a wildfire disaster risk based on an interaction relationship between a wildfire and a power facility, characterized by, Comprise: S1, by using satellite remote sensing equipment, unmanned observation device and ground meteorological sensor to obtain monitoring data, the monitoring data of each source is corrected in time and space according to time stamp and geographic coordinates, and the sampling density is adjusted according to the related weight, and the monitoring data sequence is output; S2, by inputting the monitoring data sequence into the multi-channel data transmission chain to execute dynamic transmission, the node signal strength, time delay and packet loss rate are counted, the deviation of instantaneous stability index and historical stability threshold is compared, the channel switching condition is updated, and the stable transmission sequence is output; S3, by inputting the stable transmission sequence into the fault prediction and health management process, the degradation rate value is calculated by using signal variance and attenuation slope, the autoregressive prediction is executed to update the node health score, the continuous decline interval is calibrated, and the node health record table is output; S4, by inputting the node health record table into the path reconstruction process, the replacement direction is calculated according to the signal energy gradient of adjacent nodes, the node transmission weight distribution is adjusted, the interpolation compensation and topology update are executed for the mismatch interval, and the repaired data sequence is output; S5, by inputting the repaired data sequence into the power facility forest fire risk assessment process, the feature matrix is constructed according to the node health aggregation vector and its normalized node health aggregation weight calculated according to the path level index and the monitoring data, the coupling result of the difference value of forest fire heat source intensity, power load and environmental humidity is calculated, and the comprehensive risk assessment result is output. 2.The wildfire disaster risk assessment method based on the interaction between wildfire and power facilities according to claim 1, wherein: In S1, it also includes obtaining the ground temperature distribution and heat radiation intensity data of fire source by using satellite remote sensing equipment, obtaining the appearance thermal imaging and smoke plume diffusion image data of power facilities by using unmanned observation device, and obtaining the ground temperature, humidity and wind speed data by using ground meteorological sensor, collecting the data at the same time and forming the initial monitoring data set; The initial monitoring data set is sorted according to the collection time, the average time difference of adjacent sampling intervals is calculated and linear interpolation is performed to correct the sampling period difference of different equipment, and the space alignment is performed according to the geographic coordinates to generate the monitoring data matrix under the unified time and space reference; By calculating the correlation coefficient of ground temperature, humidity and wind speed in the monitoring data matrix, the correlation coefficient is used as the sampling adjustment factor to perform weighted adjustment on the sampling point spacing, so that the sampling points in the high correlation area are kept dense, and the sampling points in the low correlation area are moderately sparse, and the time and space balanced monitoring data distribution is obtained; The monitoring data distribution is extracted in sequence according to the time index, the continuous sampling point sequence is extracted, the monitoring data of ground temperature, humidity and wind speed are respectively extracted in sequence according to the collection time index, the change amount of observation value of adjacent sampling points in the time sequence is divided by the sampling time interval, the change rate of each physical quantity is calculated, the change rate result is fitted as a continuous function according to the time index, the corresponding time trend curve is obtained, the time trend curve is used as the interpolation reference to perform data completion on the missing sampling points, and the continuity of the data is tested through residual constraint, and the monitoring data sequence after time correction, space registration and density optimization is output.

3. The method of claim 2, wherein the method further comprises: In S2, the monitoring data sequence is also input into a multi-channel data transmission chain, parallel transmission paths are established between satellite links, ground optical fiber links and wireless relay nodes, and the signal strength, transmission delay and packet loss number of each node are collected in each transmission cycle to form a node operation state table; The node operation state table is subjected to recursive calculation according to a fixed time window, in each time window, the signal strength change rate is calculated by dividing the difference between the signal strength at the current time and the signal strength at the previous window by the time interval of the two windows, and the delay drift rate is calculated by dividing the difference between the average delay at the current time and the average delay at the previous window by the time interval of the two windows; The signal strength change rate and delay drift rate of all nodes are normalized to generate an instantaneous stability index, which is compared with the historical stability threshold of the corresponding node; When the node signal strength is less than the preset signal threshold and the delay drift rate is greater than the preset drift threshold, the node is marked as an unstable node; If the above conditions are not met, the node weight remains unchanged and the node operation state table is returned to continue monitoring.

4. The method of claim 3, wherein the method further comprises: In S2, when a node is marked as an unstable node, the weighted average of its signal strength decay rate, packet loss rate and delay drift rate is calculated to generate a node risk score; When the node risk score exceeds the channel switching threshold and the packet loss rate is higher than the preset packet loss threshold, the path rearrangement judgment process is entered; if any condition is not met, the instantaneous stability index calculation process is re-executed for recursive calculation; In the path rearrangement judgment process, a neighborhood candidate node set is constructed based on the node operation state table, the neighborhood candidate node set includes adjacent nodes that have direct communication links with the unstable node; the signal strength, average delay and packet loss rate of the neighborhood candidate nodes are calculated comprehensively, and the node transmission weight is generated according to the weighted results of the node signal strength and the inverse of the delay, and the inverse of the packet loss rate; The node stability score is calculated based on the node transmission weight and the signal strength; When the stability score of any node in the neighborhood candidate node set is greater than the risk score of the unstable node, the node corresponding to the upper limit of the stability score is selected as the replacement node, the channel switching is performed and the transmission path structure is updated; if the condition is not met, the original channel configuration is maintained and the unstable node is marked as a continuous monitoring node; In the three consecutive transmission cycles after channel switching, when the node signal strength is greater than the preset signal threshold and the delay drift rate is less than the preset drift threshold, the unstable label is removed and the node operation state table is returned to execute the next cycle statistics; If any condition is not met, the node risk score calculation process is continued to be executed until the node state is restored or replaced, and finally the stable transmission sequence after dynamic adjustment is output.

5. The method of claim 4, wherein the method further comprises: In S3, the stable transmission sequence is segmented according to the node and time sequence to form the signal strength sequence, transmission delay sequence and packet loss rate sequence of each node; The signal strength sequence is extracted according to a fixed time window, the signal strength difference between adjacent sampling points is calculated and divided by the sampling time interval to obtain the signal strength change rate; calculating an average value of the signal strength change rate sequence, and averaging the square of the difference between each signal strength change rate and the average value to obtain a signal change variance; performing least squares fitting with time index as independent variable and signal strength as dependent variable, and taking the slope of the fitted straight line as signal attenuation slope; generating a node degradation rate value by normalizing the signal change variance and the signal attenuation slope; when the node degradation rate value is greater than a preset degradation threshold, determining that the node is a performance degradation node; when the node degradation rate value is less than or equal to the preset degradation threshold, keeping the node state unchanged, and returning the signal strength sequence, transmission delay sequence and packet loss rate sequence of the node to continue performing difference and fitting calculation.

6. The method of claim 5, wherein the method further comprises: In S3, further comprising: arranging the signal change variance, signal attenuation slope and time delay drift rate of the performance degradation node according to time index to form a prediction input sequence, performing autoregressive calculation on the prediction input sequence, and solving node health score prediction values of future continuous transmission periods; and calculating a health score offset by the difference between the current node health score and the node health score prediction value; when the health score offset maintains negative growth and the absolute value exceeds a preset health decline threshold in continuous three prediction periods, marking the node as a continuous decline interval; if the above conditions are not met at the same time, returning to recalculate the node degradation rate value; as an alternative condition, when the health score offset is negative growth in continuous two prediction periods and the average decline amplitude is greater than the preset health decline threshold, determining that the node is in a continuous decline interval; if any condition is not met, returning to perform autoregressive prediction operation to update the health score prediction value; when the node is determined to be in a continuous decline interval, recording the corresponding relationship between the node health score, the node degradation rate value, the signal change variance and the continuous decline interval as a node health record table, and outputting the node health record table.

7. The method of claim 6, wherein the method further comprises: In S4, further comprising: inputting the node health record table according to node number and time index to reconstruct the path, extracting the performance degradation node and its corresponding signal strength, transmission delay and packet loss rate data from the node health record table; determining the node with a health score higher than a preset health threshold and a signal strength greater than a signal threshold as a candidate healthy node, and forming a performance degradation node set and a candidate healthy node set respectively; for each performance degradation node in the performance degradation node set, calculating the signal strength difference between the candidate healthy node and the performance degradation node divided by the distance between the nodes to obtain the adjacent node signal energy gradient; when the adjacent node signal energy gradient is greater than a preset energy gradient threshold, determining that the direction is a replacement direction; if the adjacent node signal energy gradient is less than or equal to the energy gradient threshold, keeping the original path connection and returning to the node health record table to recalculate the signal energy gradient.

8. The method of claim 7, wherein the method further comprises: In S4, further comprising: after determining the replacement direction, performing node transmission capacity calculation according to the signal strength, transmission delay and packet loss rate of the candidate healthy node, and defining the node transmission capacity as a weighted function of the signal strength and the inverse of the delay, and the inverse of the packet loss rate; According to the joint result of the node transmission capability and the signal energy gradient, the node transmission weight adjustment is performed, the transmission weight of the performance degradation node is decreased by the energy gradient proportion, the transmission weight of the candidate healthy node is increased by the same proportion, and an updated node transmission weight distribution table is generated; The interpolation compensation is performed on the time section with data gap in the node transmission weight distribution table, the interpolation point value is calculated by using the time index and the signal strength change rate of the candidate healthy node, and the path topology structure is updated based on the interpolation result; The average signal strength is obtained by calculating the difference between the average signal strength of the updated path in the same time interval and the average signal strength before the update, and when the average signal strength of the updated path is greater than the preset topology stability threshold, the repair data sequence is output; if the average signal strength is less than or equal to the topology stability threshold, the node transmission weight adjustment step is returned to continue to perform the alternative direction optimization.

9. The method of claim 8, wherein the method further comprises: In S5, the repair data sequence is also input into the power facility wildfire risk assessment process, the health score, signal strength and transmission delay data of each node are extracted, and the hierarchical weighting calculation is performed according to the path level index of the node in the transmission chain topology; The nodes in the transmission chain are sorted according to the path level index, and the hierarchical index value is used as the node hierarchical weight factor; When the hierarchical index value is less than the preset hierarchical threshold, it is defined as an upper layer node, and when the hierarchical index value is greater than or equal to the threshold, it is defined as a lower layer node; The weighting coefficient is determined according to the reciprocal function of the hierarchical weight factor, the weighted accumulation is performed on the health score of each node, the health aggregation vector is generated, and the health aggregation vector is normalized to obtain the node health aggregation weight; The health aggregation vector and the ground temperature, fire source heat radiation intensity, power facility load and environmental humidity data in the monitoring data sequence are aligned according to the time index, the spatial registration is performed according to the geographical coordinates corresponding to the nodes, and a feature matrix composed of the node health aggregation vector component and the ground temperature, fire source heat radiation intensity, power facility load and environmental humidity difference value environmental monitoring data is constructed; In the feature matrix, the node health aggregation weight is used as the environmental monitoring data correction coefficient, and the influence proportion of the fire source heat radiation intensity, power load and environmental humidity difference value is weighted and corrected; The corrected environmental monitoring data is used as the input to perform the coupling calculation between the fire source heat radiation intensity, power load and environmental humidity difference value, and the coupling relationship index is obtained.

10. The method of claim 9, wherein the method further comprises: In S5, when the coupling relationship index is greater than the preset risk threshold, the time interval is marked as a high risk interval; If the coupling relationship index is less than or equal to the risk threshold, it is marked as a low risk interval; The proportion of the number of high risk intervals to the total number of evaluation intervals is counted in the continuous time window, and the mean difference value of the node health aggregation vector in the adjacent time window is divided by the time interval to obtain the node health score change rate; The high risk interval proportion and the node health score change rate are linearly combined according to the set weight coefficient to generate a comprehensive risk value, and the comprehensive risk assessment result is output. When the comprehensive risk value keeps rising trend in three continuous time windows and the proportion of high risk interval exceeds the set threshold, the power facility health early warning mechanism is triggered. If the conditions are not met, the feature matrix updating process is re-executed to correct the evaluation results.

Images