Fire risk intelligent assessment method and system for power distribution line in forest and grass region

Abstract

The present application provides a kind of power distribution line fire risk intelligent evaluation method and system for forest and grass area, it is related to computer technology field, obtains the vegetation canopy multispectral image frame sequence and environmental temperature and humidity sensing sequence continuously collected along the line direction in power distribution line overhead passage as synchronous acquisition data packet, respectively carries out vegetation canopy water stress evolution analysis and overhead conductor near-ground environment dryness trend tracking, obtains vegetation combustible state evolution mark sequence and conductor periphery dryness tendency mark sequence, input fire risk space-time alignment superposition process from this, generate fire risk alignment superposition relationship record set;Fire risk coupling trigger boundary extraction processing is carried out to the fire risk alignment superposition relationship record set, generates the fire risk disposal strategy containing firebreak opening position instruction and patrol period window planning.The present application realizes the space-time coupling evaluation and disposal decision of power distribution line corridor fire risk.

Description

Technical Field

[0001] This invention relates to the field of computer technology, and more specifically, to an intelligent assessment method and system for fire risk assessment of power distribution lines in forest and grassland areas. Background Technology

[0002] In fire prevention management of power distribution lines in forest and grassland areas, existing technologies typically employ fixed temperature sensors or infrared thermal imaging equipment to monitor the local temperature of overhead conductor corridors, or periodically acquire vegetation indices through satellite remote sensing to assess the regional fire risk level. Of these methods, fixed sensors have extremely limited monitoring range, covering only a small area directly beneath the conductors, and their deployment density is constrained by terrain and cost, making full coverage along the power distribution line corridor difficult. While satellite remote sensing offers wide coverage, its revisit period is typically several days to weeks, failing to capture the dynamic evolution of vegetation water stress over short timescales. Furthermore, the spatial resolution of satellite remote sensing is insufficient to accurately distinguish the vegetation canopy status at different locations within the overhead power distribution line corridor. Furthermore, existing technologies for fire risk assessment typically treat vegetation dryness and ambient humidity as two independent indicators, setting thresholds for each and then using simple AND / OR / OR combinations for judgment. This fails to consider the temporal coupling and alignment between the transition of vegetation canopy from stable fluctuations to continuous decline and the trend of ambient relative humidity from saturation to continuous decrease. This leads to situations where vegetation has entered a continuous drying trend but ambient humidity has not yet reached the warning level, or ambient humidity has significantly decreased but vegetation canopy remains within normal fluctuation ranges. In such cases, existing methods may generate numerous false alarms and may miss truly high-risk periods due to the misalignment of the judgment windows for the two independent indicators. Simultaneously, when generating response strategies, existing technologies typically only output fire risk levels or alarm signals, without addressing the specific locations of firebreaks in power distribution corridors or the detailed planning of patrol time windows. This makes it difficult to directly guide frontline patrol personnel in conducting precise prevention and control under limited resource conditions. Summary of the Invention

[0003] In view of the aforementioned problems, and in conjunction with the first aspect of the present invention, embodiments of the present invention provide an intelligent assessment method for fire risk of power distribution lines in forest and grassland areas, the method comprising: Acquire synchronous data packets of the forest and grassland area to be evaluated within the overhead power line channel. The synchronous data packets include a sequence of multispectral images of vegetation canopy continuously acquired along the power line and a sequence of environmental temperature and humidity sensing data within the overhead power line channel. The synchronously acquired data packets are subjected to vegetation canopy water stress evolution analysis processing to obtain vegetation combustible state evolution marker sequence. The vegetation combustible state evolution marker sequence is used to record the transition frame time marker of the normalized vegetation index from a stable fluctuation state to a continuous decay state in the vegetation canopy multispectral image frame and the frame-by-frame decay amplitude of the normalized vegetation index during the continuous decay period. The synchronously acquired data packets are processed to track the drying trend of the near-ground environment of the overhead conductor, and a drying trend marker sequence around the conductor is obtained. The drying trend marker sequence around the conductor is used to record the transition time marker when the relative humidity of the environment in the overhead conductor channel changes from a saturated humid state to a continuously decreasing state, and the time-by-time decrease rate of the relative humidity of the environment in the overhead conductor channel during the continuous decreasing period. The vegetation combustible state evolution marker sequence and the conductor perimeter drying tendency marker sequence are input into the fire hazard spatiotemporal alignment superposition process to generate a fire hazard alignment superposition relationship record set. The fire hazard alignment superposition relationship record set is used to characterize the corresponding arrangement relationship between the vegetation combustible state marker and the conductor perimeter drying state marker at the same spatial location under the same time marker. The fire hazard alignment and superposition relationship record set is processed by fire hazard coupling trigger boundary extraction to obtain fire hazard warning trigger criterion data for power distribution corridors. Based on the fire hazard warning trigger criterion data for power distribution corridors, a fire risk management strategy including guidance on the location of fire isolation zones and patrol time window planning is generated.

[0004] Furthermore, embodiments of the present invention also provide an intelligent assessment system for fire risk of power distribution lines in forest and grassland areas, comprising: A processor; a machine-readable storage medium for storing machine-executable instructions of the processor; wherein the processor is configured to execute the above-described intelligent assessment method for fire risk of power distribution lines in forest and grassland areas by executing the machine-executable instructions.

[0005] Based on the above, by combining the multispectral image frame sequence of vegetation canopy continuously acquired along the power distribution line with the temperature and humidity sensing sequence of the environment within the overhead power line tunnel as a synchronously acquired data packet, and without relying on the threshold trigger of a single indicator, this method achieves precise alignment of the state transitions of the two types of risk factors in the time dimension by extracting the transition frame time markers of the normalized vegetation index of the vegetation canopy from a stable fluctuating state to a continuously declining state, and the transition acquisition time markers of the relative humidity within the overhead power line tunnel from a saturated humid state to a continuously decreasing state. Furthermore, by using a fire risk alignment and superposition relationship record set, the flammability state of vegetation and the dryness state around the power line at the same spatial location under the same time marker are arranged and characterized, thus enabling precise alignment of the two types of risk factors in the time dimension. The extraction of the fire hazard coupling trigger boundary can simultaneously consider the synchronization degree of the two types of factors at the state transition time point and the matching degree of evolution rate during the continuous deterioration stage, thereby eliminating the false alarm and missed alarm problems caused by the misalignment of the independent judgment windows of the two types of indicators. Based on this, the generated fire hazard warning trigger criterion data for power distribution corridors is directly linked to the location guidance for firebreak opening and the planning of patrol time windows. This enables the final output fire risk management strategy to accurately locate the specific location where firebreaks need to be opened based on the distribution characteristics of the fire hazard coupling trigger boundary in the power distribution corridor space, and to plan the most effective patrol time based on the time window characteristics of fire hazard coupling triggering, significantly improving the accuracy of power distribution fire prevention management and resource utilization efficiency in forest and grassland areas. Attached Figure Description

[0006] Figure 1 This is a schematic diagram of the execution flow of the intelligent assessment method for fire risk of power distribution lines in forest and grassland areas provided in this embodiment of the invention.

[0007] Figure 2 This is a schematic diagram of the data acquisition and dual-channel parallel feature extraction process provided in an embodiment of the present invention.

[0008] Figure 3 This is a schematic diagram of the process for fire hazard spatiotemporal alignment superposition and coupling boundary extraction provided in an embodiment of the present invention.

[0009] Figure 4 This is a schematic diagram of the process for generating fire risk management strategies provided in this embodiment of the invention.

[0010] Figure 5 This is a schematic diagram comparing the state space trajectory of the evolution of a bivariate fire risk situation, provided in an embodiment of the present invention.

[0011] Figure 6 This is a schematic diagram illustrating the joint analysis of vegetation attenuation pulse events and dryness fluctuation rhythms provided in an embodiment of the present invention.

[0012] Figure 7This is a schematic diagram illustrating spatial clustering and key indicator comparison of fire hazard composite risk areas in power distribution corridors, provided in an embodiment of the present invention. Detailed Implementation

[0013] Figure 1 This is a flowchart illustrating an intelligent fire risk assessment method for power distribution lines in forest and grassland areas, provided by one embodiment of the present invention. Co-located data collection points are deployed along the overhead power line corridors to simultaneously collect multispectral images of the vegetation canopy and environmental temperature and humidity data within the overhead conductor corridors. By independently analyzing and spatiotemporally overlaying the evolution of vegetation canopy water stress and the drying trend of the near-ground environment of the conductors, the fire hazard coupling trigger boundary is extracted, generating a fire risk management strategy that includes guidance on the location of firebreaks and patrol time window planning. All data collection in this embodiment complies with forestry and power industry data collection standards and does not involve personal privacy information.

[0014] Step S110: Obtain synchronous acquisition data packets of the forest and grassland area to be evaluated within the overhead power line channel. The synchronous acquisition data packets include a sequence of multispectral images of the vegetation canopy continuously acquired along the power line and a sequence of environmental temperature and humidity sensors within the overhead power line channel.

[0015] Step S111: Along the overhead power line corridor, a co-located data acquisition point array is set up at a preset spatial interval. Each co-located data acquisition point is equipped with both a multispectral image acquisition device and an environmental temperature and humidity sensor.

[0016] The preset spatial intervals are determined based on the span of the power distribution line corridor and the terrain undulations, using equal intervals between the spans. The lens of the multispectral image acquisition equipment is fixedly mounted on the crossarm of the tower directly above the overhead power line or on a dedicated bracket, with a top-down view, ensuring that the imaging range covers the vegetation canopy of the forest and grassland area below the power line. The sensor probes of the environmental temperature and humidity sensing equipment are deployed in the near-ground space within the overhead power line corridor, with the installation height determined by a preset proportion of the distance between the power line and the ground. The co-located acquisition points are numbered incrementally along the power distribution line route.

[0017] Step S112: Drive the multispectral image acquisition device in each co-located acquisition point to take pictures of the vegetation canopy in the forest and grassland area below the power distribution overhead channel from a top-down direction, and obtain the vegetation canopy multispectral image frame corresponding to the co-located acquisition point. The vegetation canopy multispectral image frame includes red band reflectance value and near-infrared band reflectance value.

[0018] The multispectral image acquisition device simultaneously acquires red band and near-infrared band images in each acquisition cycle. Each pixel in the red band image stores the red band reflectance value R at that pixel location, and each pixel in the near-infrared band image stores the near-infrared band reflectance value NIR at that pixel location. The vegetation canopy multispectral image frame is a multi-layer two-dimensional array structure, with the first channel storing the R value and the second channel storing the NIR value. The frame header records the acquisition time and the co-located acquisition point number.

[0019] Step S113: Arrange the vegetation canopy multispectral image frames acquired at each co-located acquisition point at multiple consecutive acquisition times in chronological order of acquisition time to form a vegetation canopy multispectral image frame fragment corresponding to that co-located acquisition point.

[0020] For each co-located acquisition point, a frame sequence buffer is established. After the multispectral image acquisition equipment completes its shooting within each acquisition cycle, the generated vegetation canopy multispectral image frames are appended to the frame sequence buffer of that co-located acquisition point. The vegetation canopy multispectral image frames in the frame sequence buffer are arranged in ascending order according to the acquisition time, forming a vegetation canopy multispectral image frame segment.

[0021] Step S114: Drive the environmental temperature and humidity sensing device in each co-located acquisition point to sense and measure the relative humidity of the air environment in the near-ground space within the overhead conductor channel, obtain the relative humidity value of the overhead conductor channel corresponding to the co-located acquisition point, and arrange the relative humidity values ​​of the overhead conductor channel obtained by each co-located acquisition point at multiple consecutive acquisition times in chronological order of acquisition time to form the relative humidity segment of the overhead conductor channel corresponding to the co-located acquisition point.

[0022] The environmental temperature and humidity sensor measures the relative humidity of the near-ground air environment within the overhead power line tunnel once per data acquisition cycle, outputting a relative humidity value H in percentage form. The relative humidity value H, along with the acquisition time and the co-located acquisition point number, is recorded in the humidity data buffer. The relative humidity values ​​in the humidity data buffer are arranged in ascending order of acquisition time, forming a relative humidity segment within the overhead power line tunnel.

[0023] Step S115: Establish a correlation between the acquisition time of each multispectral image frame in the vegetation canopy multispectral image frame segment corresponding to the same co-located acquisition point and the relative humidity value of the environment in the overhead conductor channel at the same acquisition time, generate a corresponding record of the image frame and the environmental humidity of the co-located acquisition point, arrange the corresponding records of the image frame and the environmental humidity of all co-located acquisition points according to the spatial order of the power distribution line, and combine them into a vegetation canopy multispectral image frame sequence and an environmental temperature and humidity sensing sequence in the overhead conductor channel.

[0024] Using the acquisition time as the primary key, isolinking is performed on vegetation canopy multispectral image frame fragments and relative humidity fragments within the overhead conductor tunnel at the same co-located acquisition point. A corresponding record is generated for each acquisition time, containing the co-located acquisition point number, acquisition time, the storage address of the vegetation canopy multispectral image frame at that time, and the relative humidity value H at that time. All corresponding records for co-located acquisition points are sorted in ascending order by co-located acquisition point number, and the corresponding records within each co-located acquisition point are sorted in ascending order by acquisition time. The vegetation canopy multispectral image frames from all corresponding records are extracted and stitched together to form a vegetation canopy multispectral image frame sequence, and the relative humidity values ​​H from all corresponding records are extracted and stitched together to form an environmental temperature and humidity sensing sequence within the overhead conductor tunnel.

[0025] Step S116: Package the vegetation canopy multispectral image frame sequence and the environmental temperature and humidity sensing sequence within the overhead conductor channel into a synchronous acquisition data package.

[0026] The synchronous acquisition data packet contains two independent data blocks: a frame sequence data area and a humidity sequence data area, as well as a mapping index area. The mapping index area records the bidirectional mapping relationship between the storage offset address of each frame in the frame sequence data area and the storage offset address of the relative humidity value at the same acquisition time in the humidity sequence data area. The header of the synchronous acquisition data packet records the total number of co-located acquisition points, the total acquisition time span, and the preset spatial interval parameters.

[0027] Step S120: Perform vegetation canopy water stress evolution analysis on the synchronously acquired data packet to obtain a vegetation combustible state evolution marker sequence. The vegetation combustible state evolution marker sequence is used to record the transition frame time marker of the normalized vegetation index from a stable fluctuation state to a continuous decay state in the vegetation canopy multispectral image frame, and the frame-by-frame decay amplitude of the normalized vegetation index during the continuous decay period.

[0028] Step S121: Extract the vegetation canopy multispectral image frame fragments corresponding to each co-located acquisition point from the synchronous acquisition data packet. Extract the red band reflectance value and near-infrared band reflectance value for each multispectral image frame in the vegetation canopy multispectral image frame fragment. Calculate the normalized vegetation index based on the red band reflectance value and near-infrared band reflectance value of each multispectral image frame. Use the normalized vegetation index as the vegetation canopy greenness characterization of that multispectral image frame.

[0029] Extract the corresponding vegetation canopy multispectral image frame segments from the frame sequence data area of ​​the synchronously acquired data packets, according to the co-located acquisition point number. For each frame in the segment, read the red band reflectance value R and the near-infrared band reflectance value NIR, and calculate the Normalized Difference Vegetation Index (NDVI) = (NIR - R) / (NIR + R + ε), where ε is a very small positive number to prevent the denominator from being zero. The NDVI value ranges from -1 to 1, and when the vegetation is dense and healthy, the NDVI is close to 1. The NDVI value of this frame is used as a characterization of the vegetation canopy greenness and assigned to the greenness field of this frame.

[0030] Step S122: Arrange the vegetation canopy greenness characterization of each multispectral image frame in the vegetation canopy multispectral image frame segment corresponding to the same co-located acquisition point according to the order of acquisition time to generate a time series of vegetation canopy greenness characterization. Perform stationarity transition detection processing on the time series of vegetation canopy greenness characterization and extract the transition point acquisition time when the vegetation canopy greenness characterization changes from a stable fluctuation state with a fluctuation amplitude less than a preset stationarity threshold to a continuous monotonically decreasing state. Record the transition point acquisition time as the transition frame time marker when the normalized vegetation index changes from a stable fluctuation state to a continuous decay state.

[0031] The NDVI values ​​of all frames from the same co-located sampling point are arranged in ascending order of sampling time to form a time series representing the vegetation canopy greenness. A sliding window segmented stationarity detection algorithm is used to process this time series. A preset window length is used to slide across the time series, and the standard deviation of the NDVI values ​​within each window is calculated as the fluctuation amplitude. When the fluctuation amplitude within a window is less than a preset stationarity threshold, the window is marked as a stable fluctuation window. Along the time axis, when the time series first shows a strict decrease in NDVI values ​​for K1 consecutive sampling times after the stable fluctuation window, the first of these K1 consecutive sampling times is marked as the inflection point sampling time. This inflection point sampling time is the time marker of the transition frame from a stable fluctuation state to a continuously declining state in the Normalized Difference Vegetation Index (NDVI), where K1 is the preset number of decrease determinations. This inflection frame time marker stores the sampling time value Tndvi corresponding to the inflection point.

[0032] Step S123: Starting from the transition frame time mark, extract the vegetation canopy greenness characterization value for each subsequent acquisition time in the vegetation canopy greenness characterization value time series along the acquisition time direction. Calculate the difference between the vegetation canopy greenness characterization values ​​of two adjacent acquisition times and record the difference as the frame-by-frame attenuation amount. Arrange the frame-by-frame attenuation amounts corresponding to each acquisition time in the continuous period after the transition frame time mark according to the order of acquisition time to form a frame-by-frame attenuation amount sequence. Record the frame-by-frame attenuation amount sequence as the frame-by-frame attenuation amplitude of the normalized vegetation index during the continuous attenuation period.

[0033] All elements with a sampling time greater than or equal to Tndvi are extracted from the time series of vegetation canopy greenness indicators to form a continuous decay period sequence. For each sampling time k in this sequence, the difference between the NDVI value at that time and the NDVI value at the previous time is calculated as ΔNDVIk = NDVIk-1 - NDVIk. All ΔNDVIk values ​​are arranged in ascending order of sampling time to form a frame-by-frame decay sequence, which represents the frame-by-frame decay magnitude of the Normalized Difference Vegetation Index (NDVI) during the continuous decay period.

[0034] Step S124: Combine the transition frame time marker of the normalized vegetation index (NVC) from a stable fluctuation state to a continuous decay state corresponding to each co-located collection point with the frame-by-frame decay amplitude of the NVC during the continuous decay period to form the vegetation combustible state evolution marker record for that co-located collection point. Summarize the vegetation combustible state evolution marker records of all co-located collection points according to their spatial location to generate a vegetation combustible state evolution marker sequence.

[0035] For each co-located sampling point, a vegetation combustible state evolution marker record data structure is created, containing a co-located sampling point number field, a transition frame time marker field, and a frame-by-frame attenuation amplitude sequence field. All vegetation combustible state evolution marker records from all co-located sampling points are arranged in ascending order by co-located sampling point number, forming a vegetation combustible state evolution marker sequence.

[0036] Step S130: Perform near-ground environment drying trend tracking processing on the synchronously acquired data packets to obtain a drying tendency marker sequence around the conductor. The drying tendency marker sequence around the conductor is used to record the transition time marker when the relative humidity in the overhead conductor channel changes from a saturated humid state to a continuously decreasing state, and the hourly decrease rate of the relative humidity in the overhead conductor channel during the continuously decreasing period.

[0037] Step S131: Extract the relative humidity segment of the overhead conductor channel corresponding to each co-located acquisition point from the synchronous acquisition data packet, and arrange the relative humidity values ​​of the overhead conductor channel in the relative humidity segment of the overhead conductor channel according to the order of acquisition time to form the time series of relative humidity of the overhead conductor channel.

[0038] Extract the corresponding relative humidity segments within the overhead conductor tunnel from the humidity sequence data area of ​​the synchronously acquired data packets, according to the co-located acquisition point number. Arrange the relative humidity values ​​H in the segments in ascending order of acquisition time to form a time series of relative humidity within the overhead conductor tunnel.

[0039] Step S132: Perform saturated humidity state identification processing on the time series of relative humidity in the overhead conductor channel, extract continuous segments in the overhead conductor channel where the relative humidity value is continuously higher than the preset saturated humidity threshold, and record the last collection time in the continuous segment as the saturated humidity state termination time.

[0040] The relative humidity time series within the overhead conductor tunnel is traversed, and consecutive sampling time segments where the relative humidity value H is greater than the preset saturation humidity threshold Hsat are identified as saturated humidification state segments. The last sampling time in a saturated humidification state segment is recorded as the saturated humidification state termination time Tsat.

[0041] Step S133: Starting from the next acquisition time after the end of the saturated humid state, track the time series of relative humidity in the overhead conductor channel backward along the acquisition time direction. Record the acquisition time when the relative humidity value in the overhead conductor channel first falls below the preset drying trigger threshold and the relative humidity value in the overhead conductor channel remains below the preset drying trigger threshold for multiple consecutive acquisition times as the transition acquisition time mark from the saturated humid state to the continuously decreasing state.

[0042] Starting from the first acquisition time after Tsat, the system tracks the first acquisition time where the relative humidity value H is lower than the preset dryness trigger threshold Hdry, and the relative humidity value H is lower than Hdry for the next K2 consecutive acquisition times. This acquisition time is recorded as the transition acquisition time marker Tdry, where K2 is the preset number of consecutive judgments. Tdry represents the transition time node when the near-ground air environment in the overhead conductor channel officially changes from a saturated and humid state to a continuously dry and declining state.

[0043] Step S134: Starting from the turning point acquisition time mark, extract the relative humidity value of the overhead conductor channel at each subsequent acquisition time in the time series of relative humidity of the overhead conductor channel along the acquisition time direction, and calculate the decrease difference between the relative humidity value of the overhead conductor channel at each acquisition time and the relative humidity value of the overhead conductor channel at the previous acquisition time.

[0044] Starting from Tdry, for each subsequent acquisition time k in the relative humidity time series within the overhead conductor channel, calculate the decrease difference ΔHk = Hk-1 - Hk. When Hk is greater than Hk-1, i.e., the relative humidity temporarily rises, ΔHk takes a zero value, indicating that the decrease process is interrupted but does not affect the continuity of the recorded decrease difference sequence.

[0045] Step S135: Arrange the decrease difference corresponding to each acquisition time in the continuous period after the turning acquisition time mark according to the order of acquisition time to form a time-by-time decrease difference sequence. Record the time-by-time decrease difference sequence as the time-by-time decrease rate of the relative humidity of the environment in the overhead conductor channel during the continuous decrease period.

[0046] All ΔHk values ​​are arranged in ascending order according to the acquisition time to form a time-by-time decreasing difference sequence. This time-by-time decreasing difference sequence is the time-by-time decreasing rate of relative humidity in the overhead conductor channel during the continuous decreasing period.

[0047] Step S136: Combine the transition time markers from saturated humid state to continuously decreasing state corresponding to each co-located acquisition point with the hourly decrease rate of relative humidity in the overhead conductor channel during the continuously decreasing period to form a conductor perimeter drying tendency marker record for that co-located acquisition point. Summarize all conductor perimeter drying tendency marker records for all co-located acquisition points according to spatial location to generate a conductor perimeter drying tendency marker sequence.

[0048] For each co-located acquisition point, create a data structure for marking the drying tendency around the conductor, including a co-located acquisition point number field, a turning acquisition time marker field, and a time-by-time descent rate sequence field. Sort all the records for marking the drying tendency around the conductor for all co-located acquisition points in ascending order by co-located acquisition point number to form a sequence of drying tendency markings around the conductor.

[0049] Step S140: Input the vegetation combustible state evolution marker sequence and the conductor perimeter dryness tendency marker sequence into the fire hazard spatiotemporal alignment superposition processing process to generate a fire hazard alignment superposition relationship record set. The fire hazard alignment superposition relationship record set is used to characterize the corresponding arrangement relationship between the vegetation combustible state marker and the conductor perimeter dryness state marker at the same spatial location under the same time marker.

[0050] Step S141: Extract the transition frame time marker from the stable fluctuation state to the continuous decay state of the normalized vegetation index of each co-located collection point from the vegetation combustible state evolution marker sequence, and use the transition frame time marker as the time node when the vegetation of the co-located collection point enters the combustible state.

[0051] Traverse the vegetation combustible state evolution marker records of each co-located collection point in the vegetation combustible state evolution marker sequence, read the transition frame time marker field value Tndvi, and use this value as the time node Tveg when the vegetation at that co-located collection point enters the combustible state.

[0052] Step S142: Extract the transition acquisition time marker from the saturated wet state to the continuously decreasing state of each co-located acquisition point from the drying tendency marker sequence around the conductor, and use the transition acquisition time marker as the time node when the conductor periphery of the co-located acquisition point enters the dry state.

[0053] Traverse the drying tendency marker records around each co-located acquisition point in the drying tendency marker sequence around the conductor, read the turning acquisition time marker field value Tdry, and use this value as the time node Twire at which the conductor around the co-located acquisition point enters the drying state.

[0054] Step S143: Compare the time nodes when the vegetation at the same co-located collection point enters the combustible state with the time nodes when the area around the conductor enters the dry state. When the time interval between the time nodes when the vegetation enters the combustible state and the time nodes when the area around the conductor enters the dry state is less than the preset fire hazard coupling judgment time window, mark the co-located collection point as a fire hazard time coupling point.

[0055] Calculate the absolute value of the time difference between the co-located acquisition points, ΔTcp = |Tveg - Twire|. Compare ΔTcp with the preset fire hazard coupling judgment time window, Wcp. If ΔTcp ≤ Wcp, the co-located acquisition point is marked as a fire hazard time coupling point, indicating that the vegetation entering the combustible state and the area around the conductor entering the dry state occur almost synchronously in time, with the highest degree of fire hazard coupling.

[0056] Step S144: Compare the time nodes when the vegetation at the same co-located collection point enters the combustible state with the time nodes when the surrounding area of ​​the conductor enters the dry state. When the time interval between the time node when the vegetation enters the combustible state and the time node when the surrounding area of ​​the conductor enters the dry state is greater than the preset fire hazard coupling judgment time window and less than the preset fire hazard proximity judgment time window, mark the co-located collection point as a fire hazard proximity point.

[0057] If Wcp < ΔTcp ≤ Wadj, where Wadj is the preset fire hazard proximity determination time window and Wadj > Wcp, then the co-located collection point is marked as a fire hazard proximity point, indicating that there is a certain time interval between the vegetation entering the combustible state and the area around the conductor entering the dry state, but they have not yet moved far apart.

[0058] Step S145: Compare the time nodes when the vegetation at the same co-located collection point enters the combustible state with the time nodes when the area around the conductor enters the dry state. If the time interval between the time nodes when the vegetation enters the combustible state and the time nodes when the area around the conductor enters the dry state is greater than the preset fire hazard proximity judgment time window, mark the co-located collection point as a fire hazard time far away point.

[0059] If ΔTcp>Wadj, then the co-located collection point is marked as a point far from the fire hazard time, indicating that the time interval between the vegetation entering the combustible state and the area around the conductor entering the dry state is large, and the degree of fire hazard coupling is low.

[0060] Step S146: Associate and store the fire hazard point marker type of each co-located collection point with the spatial coordinates of the co-located collection point to form a fire hazard alignment and superposition relationship record. Summarize the fire hazard alignment and superposition relationship records of all co-located collection points into a fire hazard alignment and superposition relationship record set. The fire hazard point marker type includes fire hazard time-coupled points, fire hazard time adjacent points, and fire hazard time far away points.

[0061] For each co-located data collection point, a fire hazard alignment and overlay record is created, including a co-located data collection point number field, a spatial location coordinate field, a fire hazard point marker type field, a ΔTcp field, a Tveg field, and a Twire field. All fire hazard alignment and overlay record records for co-located data collection points are arranged in ascending order of co-located data collection point number, forming a fire hazard alignment and overlay record set.

[0062] Step S150: Perform fire hazard coupling trigger boundary extraction processing on the fire hazard alignment and superposition relationship record set to obtain fire hazard warning trigger criterion data for power distribution corridors, and generate fire risk handling strategies based on the fire hazard warning trigger criterion data for power distribution corridors, including guidance on the location of fire isolation zones and patrol time window planning.

[0063] Step S151: Extract the frame-by-frame attenuation magnitude of the normalized vegetation index and the hour-by-hour decrease rate of the relative humidity in the overhead conductor channel during the continuous attenuation period from the fire hazard alignment superposition relationship record set for all co-located acquisition points marked as fire hazard time coupling points.

[0064] The fire hazard location superposition relationship record set is traversed, and records with fire hazard location marker type of fire hazard time coupling location are selected. Based on the co-located acquisition point number of these records, the corresponding frame-by-frame attenuation amplitude sequence is extracted from the vegetation combustible state evolution marker sequence, and the corresponding time-by-time decrease rate sequence is extracted from the conductor perimeter drying tendency marker sequence.

[0065] Step S152: Extract the maximum value of the frame-by-frame attenuation amplitude of the fire time-coupled point as the upper bound reference value of the vegetation yellowing and spreading rate, and extract the minimum value of the frame-by-frame attenuation amplitude of the fire time-coupled point as the lower bound reference value of the vegetation yellowing and spreading rate.

[0066] A global comparison is performed on all ΔNDVI values ​​in the frame-by-frame attenuation amplitude sequence of all fire time-coupled points. The maximum value is taken as the upper bound reference value Vupper for the vegetation yellowing and spreading rate, and the minimum value is taken as the lower bound reference value Vlower for the vegetation yellowing and spreading rate.

[0067] Step S153: Extract the maximum value of the hourly decrease rate of the fire hazard time-coupled point as the upper limit reference value of the environmental drying rate, and extract the minimum value of the hourly decrease rate of the fire hazard time-coupled point as the lower limit reference value of the environmental drying rate.

[0068] A global comparison is performed on all ΔH values ​​in the time-by-time decrease rate sequence of all fire hazard time-coupled locations. The maximum value is taken as the upper bound reference value of the environmental drying rate, Dupper, and the minimum value is taken as the lower bound reference value of the environmental drying rate, Dlower.

[0069] Step S154: Combine the lower bound reference value of vegetation yellowing spread rate and the lower bound reference value of environmental drying rate into the lower boundary criterion for triggering fire risk warning in power distribution corridor. The lower boundary criterion for triggering fire risk warning in power distribution corridor is used to trigger a fire risk warning when the frame-by-frame attenuation amplitude of any co-located acquisition point to be monitored reaches the lower bound reference value of vegetation yellowing spread rate and its hour-by-hour decrease rate simultaneously reaches the lower bound reference value of environmental drying rate.

[0070] The lower boundary criterion for triggering a fire hazard warning in the power distribution corridor is a logical AND conditional expression: a fire hazard warning is triggered when ΔNDVI ≥ Vlower and ΔH ≥ Dlower at the monitored point. The fire hazard warning is the lowest level of fire hazard warning, indicating the need to pay attention to the dryness and flammability trends of the vegetation at that point.

[0071] Step S155: Combine the upper limit reference value of vegetation yellowing spread rate and the upper limit reference value of environmental drying rate into the upper boundary criterion for triggering fire risk warning in power distribution corridor. The upper boundary criterion for triggering fire risk warning in power distribution corridor is used to trigger an emergency fire alarm when the frame-by-frame attenuation amplitude of any co-located acquisition point to be monitored reaches the upper limit reference value of vegetation yellowing spread rate and its hourly decrease rate simultaneously reaches the upper limit reference value of environmental drying rate.

[0072] The upper boundary criterion for triggering a fire hazard warning in the power distribution corridor is a logical AND conditional expression: a fire emergency alarm is triggered when ΔNDVI ≥ Vupper and ΔH ≥ Dupper at the monitored point. The fire emergency alarm is the highest level of fire hazard warning, indicating that the fire hazard coupling at that point is extremely severe, and immediate emergency fire prevention measures are required.

[0073] Step S156: Extract the spatial coordinates of the co-located collection points marked as fire time coupling points in the fire hazard alignment and superposition relationship record set, calculate the continuous dense sections of the fire time coupling points in the spatial distribution along the overhead power distribution channel, and mark the area outside the boundary of the power distribution corridor covered by the starting coordinates of the continuous dense sections extending outward in the opposite direction of the power distribution line and the ending coordinates extending outward in the positive direction of the power distribution line as the location guide for the opening of the fire isolation zone.

[0074] The spatial coordinates of all fire hazard time-coupled points are projected one-dimensionally along the power distribution line. A density-based continuous segment detection algorithm is used to identify continuous dense segments with spatial spacing less than a preset dense spacing threshold. For each continuous dense segment, the starting point coordinates are extended outward by a preset distance in the opposite direction of the power distribution line to obtain the starting point of the firebreak, and the ending point coordinates are extended outward by a preset distance in the forward direction of the power distribution line to obtain the ending point of the firebreak. The area outside the boundary of the power distribution line corridor between the starting and ending points of the firebreak serves as the location guide for the firebreak.

[0075] Step S157: Extract the time nodes when vegetation enters the combustible state and the time nodes when the perimeter of the conductor enters the dry state from the fire hazard alignment and superposition relationship record set marked as fire hazard time coupling points. Calculate the time overlap interval between the time nodes when vegetation enters the combustible state and the time nodes when the perimeter of the conductor enters the dry state. Convert the time overlap interval into the start and end times of the patrol period window planning and generate the patrol period window planning.

[0076] For each fire hazard time-coupled point, extract Tveg and Twire, and take the later of the two as the start time Tstart of the patrol time window planning. Take Tstart plus the preset patrol duration Tdur as the end time Tend. Sort the patrol time windows of all fire hazard time-coupled points in ascending order of start time, merge windows with overlapping times, and generate the patrol time window plan.

[0077] Step S210: After acquiring the synchronous data acquisition data of the forest and grassland area to be evaluated within the overhead power line corridor, acquire the tower tilt sensing data corresponding to each co-located acquisition point along the overhead power line corridor. The tower tilt sensing data includes the offset direction angle and offset distance of the tower top relative to the tower base.

[0078] The tower tilt sensing data is obtained from the tilt sensors of the power distribution tower online monitoring system. Each co-located collection point is associated with the tilt sensor data of the nearest tower within its span. The offset direction angle is the azimuth angle of the tower top relative to the tower base in the horizontal plane, and the offset distance is the amount of offset of the tower top relative to the tower base in the horizontal plane.

[0079] Step S220: Calculate a three-dimensional displacement vector of the tower top relative to the tower base according to the offset direction angle and offset distance in the tower tilt sensing data corresponding to each co-located collection point, and at least decompose a vertical settlement component along the plumb direction and a transverse horizontal displacement component perpendicular to the trend of the distribution line; subtract the vertical settlement component from the designed clearance of the overhead conductor to the ground, so as to obtain the actual clearance value of the overhead conductor to the ground considering the influence of tower tilt and settlement.

[0080] The three-dimensional displacement vector is decomposed into three orthogonal components along the trend of the distribution line, perpendicular to the trend of the distribution line, and in the plumb direction. The vertical settlement component along the plumb direction is the projection length of the offset distance in the vertical plane, which is calculated according to the tower height and the offset distance. Subtract the vertical settlement component from the designed clearance of the overhead conductor to the ground, to obtain the actual clearance value of the overhead conductor to the ground Hactual=Hdesign-δvertical, where Hdesign is the designed clearance of the overhead conductor to the ground, and δvertical is the vertical settlement component. If the vertical settlement component is a negative value, that is, the tower sinks, the actual clearance value becomes smaller.

[0081] Step S230: Acquire vegetation canopy height data in the overhead conductor channel corresponding to each co-located collection point, perform difference calculation on the actual clearance value of the overhead conductor to the ground corrected by tower tilt and the vegetation canopy height, so as to obtain the vertical clearance distance between the overhead conductor and the top surface of the vegetation canopy.

[0082] The vegetation canopy height data is acquired through lidar scanning or stereo vision measurement, and stored as the height value above the ground of the highest point of the vegetation canopy below each co-located collection point. The vertical clearance distance Hclearance=Hactual-Hcanopy, where Hcanopy is the vegetation canopy height. A smaller Hclearance indicates a closer distance between the overhead conductor and the top surface of the vegetation canopy.

[0083] Step S240: When the vertical clearance distance is less than the preset safety clearance threshold of the conductor to the vegetation, generate a conductor vegetation obstacle hazard mark corresponding to the co-located collection point, append the conductor vegetation obstacle hazard mark to the fire risk alignment superposition relation record set, and update the fire risk alignment superposition relation record set.

[0084] The preset safety clearance threshold of the conductor to the vegetation Hsafe is determined according to the voltage level of the distribution line. If Hclearance<Hsafe, a conductor vegetation obstacle hazard mark field is added to the fire risk alignment superposition relation record of the co-located collection point and set to true. The updated fire risk alignment superposition relation record set contains both fire risk time coupling information and conductor vegetation obstacle hazard information.

[0085] Step S310: After generating the fire risk warning triggering criterion data for the power distribution corridor, obtain the historical fire record data of the forest and grassland area to be evaluated. The historical fire record data includes the geographical coordinates of each fire and the date and time of the fire.

[0086] Historical fire incident records are obtained from the forestry fire prevention management database or power operation and maintenance ledger and stored in the form of a fire incident record list. Each record includes the fire incident number, geographical coordinates (longitude and latitude), and the date and timestamp of the fire.

[0087] Step S320: Spatial matching processing is performed between the geographical coordinates of each fire occurrence in the historical fire record data and the spatial location coordinates of the co-located collection points to determine the historical fire markers corresponding to the co-located collection points where historical fires have occurred. The frame-by-frame attenuation magnitude of the historical normalized vegetation index of the co-located collection points with historical fire markers before the date and time of the fire occurrence and the hour-by-hour decrease rate of the relative humidity in the historical overhead conductor channel during the continuous decrease period are extracted.

[0088] A nearest neighbor spatial matching algorithm is used to calculate the distance between the geographic coordinates of each fire and the spatial coordinates of all co-located data collection points. The co-located data collection point with the smallest distance, which is less than a preset matching radius, is selected as the matching point for that historical fire. This co-located data collection point is assigned a historical fire marker. The historical frame-by-frame decay amplitude and historical time-by-time decay rate of this point before the date and time of the fire are extracted from the historical archived vegetation combustibility evolution marker sequence and the conductor perimeter dryness tendency marker sequence.

[0089] Step S330: Compare the historical frame-by-frame attenuation amplitude of co-located acquisition points with historical fire markers with the lower bound reference value of vegetation withering and spreading rate. When the historical frame-by-frame attenuation amplitude is lower than the lower bound reference value of vegetation withering and spreading rate, use the historical frame-by-frame attenuation amplitude as the new lower bound reference value of vegetation withering and spreading rate.

[0090] If the minimum value Vhist in the historical frame-by-frame decay rate is less than the current lower bound reference value Vlower for vegetation yellowing and spreading rate, then Vlower is updated to Vhist. This indicates that among the locations where actual fires occurred in the past, there were cases where the vegetation decay rate was slower than the current lower bound reference value. Therefore, the warning lower bound is lowered to cover actual fire risk precursors in the past.

[0091] Step S340: Compare the historical moment-by-moment decline rate of the co-located collection points with historical fire marks against the lower reference value of the environmental drying rate. When the historical moment-by-moment decline rate is lower than the lower reference value of the environmental drying rate, take the historical moment-by-moment decline rate as the new lower reference value of the environmental drying rate, and use the new lower reference value of the vegetation withering spread rate and the new lower reference value of the environmental drying rate to update the lower boundary criterion for triggering fire risk warning of the distribution line corridor.

[0092] If the minimum value Dhist in the historical moment-by-moment decline rate is less than the current lower reference value Dlower of the environmental drying rate, update Dlower to Dhist. Combine the updated Vlower and Dlower into a new lower boundary criterion for triggering fire risk warning of the distribution line corridor, and replace the original lower boundary criterion.

[0093] Step S410: After generating the vegetation flammable state evolution mark sequence, acquire the daily precipitation observation record sequence of the forest and grass area to be assessed within the same collection period, compare the daily precipitation value in the daily precipitation observation record sequence with the preset effective precipitation threshold, mark the date whose daily precipitation value is lower than the preset effective precipitation threshold as an invalid precipitation day, and mark the date whose daily precipitation value is higher than the preset effective precipitation threshold as an effective precipitation day.

[0094] The daily precipitation observation record sequence is obtained from a meteorological observation station or a rain gauge, and stored in the form of a precipitation table indexed by date. The preset effective precipitation threshold Pmin is the minimum daily precipitation that can effectively supplement the water content of the vegetation canopy. Compare the daily precipitation value P with Pmin, mark P<Pmin as an invalid precipitation day, and mark P≥Pmin as an effective precipitation day.

[0095] Step S420: Perform continuous cumulative counting processing on invalid precipitation days, reset the continuous cumulative count to zero and restart accumulation when an effective precipitation day is encountered, to obtain the cumulative number of consecutive invalid precipitation days, perform association and correspondence processing on the cumulative number of consecutive invalid precipitation days and the turning frame time mark when the normalized difference vegetation index transitions from a stable fluctuation state to a continuous attenuation state, and extract the cumulative number of consecutive invalid precipitation days corresponding to the turning frame time mark as the critical continuous drought tolerance day value of vegetation.

[0096] Traverse the daily precipitation observation record sequence, and maintain a consecutive invalid precipitation day cumulative counter. The counter increments by 1 when an invalid precipitation day is encountered, and resets to zero when an effective precipitation day is encountered. Record the count value corresponding to each date as the cumulative day value. In the vegetation flammable state evolution mark sequence, read the turning frame time mark Tndvi of each co-located collection point, look up the cumulative day value of the date corresponding to Tndvi in the daily precipitation observation record sequence, and take it as the critical continuous drought tolerance day value Ddrought of vegetation.

[0097] Step S430: Extract the total decline of the normalized vegetation index in the time series of vegetation canopy greenness characteristics during the continuous decline period, and calculate the ratio of the total decline to the cumulative number of days with continuous ineffective precipitation to obtain the vegetation greenness decline rate per unit of ineffective precipitation day.

[0098] The total attenuation, ΔNDVItotal, is the difference between the NDVI value at the time marker Tndvi (the transition frame) and the NDVI value at the end of the continuous attenuation period. The vegetation greenness attenuation rate per unit of ineffective precipitation, Rdecline = ΔNDVItotal / Ddrought, represents the average rate of vegetation greenness decrease per consecutive day without effective precipitation.

[0099] Step S440: Input the critical duration of drought resistance of vegetation and the vegetation greenness decay rate per unit of ineffective precipitation as leading parameters into the update process of the vegetation combustible state evolution marker sequence, and adjust the preset stationarity threshold for the stationarity transition detection processing in the time series of vegetation canopy greenness characterization quantities.

[0100] The critical duration of drought tolerance for vegetation (Ddrought) and the vegetation greenness decay rate (Rdecline) are used as feedback parameters. A preset stability threshold is dynamically adjusted based on Rdecline. The adjustment formula is: New stability threshold = Original stability threshold × Adjustment coefficient, where the adjustment coefficient is inversely proportional to Rdecline. A larger Rdecline indicates greater sensitivity of vegetation to drought, and the stability threshold should be lowered to detect the decline inflection point earlier.

[0101] For example, in step S510: after generating the drying tendency marker sequence around the conductor, the surface temperature sensing data of the overhead conductor corresponding to each co-located collection point along the overhead distribution channel is obtained. The surface temperature sensing data of the overhead conductor includes the temperature values ​​of the surface of the overhead conductor at multiple collection times.

[0102] The surface temperature sensing data of overhead conductors is acquired from the online monitoring system for overhead conductor temperature and stored using the co-located acquisition point number as an index. The temperature sensing data for each co-located acquisition point is a sequence of temperature values ​​collected by the corresponding overhead conductor surface temperature sensor at multiple consecutive acquisition times.

[0103] Step S520: Arrange the temperature values ​​of the overhead conductor surface at multiple acquisition times in chronological order to form a time series of overhead conductor surface temperature. Perform temperature rise trend identification processing on the time series of overhead conductor surface temperature, and extract the temperature rise start acquisition time mark when the overhead conductor surface temperature changes from the ambient temperature level to a continuous temperature rise state, and the temperature rise rate of the overhead conductor surface temperature during the continuous temperature rise period.

[0104] Arrange the temperature values in ascending order by collection time to form a time series of the surface temperature of the overhead conductor. Using a continuous trend detection method similar to that in steps S132 to S133, extract the temperature rise start collection time marker Ttemp at which the conductor surface temperature transitions from the reference ambient temperature level to a continuous temperature rise state, and take the temperature rise amount between adjacent collection times within the continuous temperature rise period as the time-by-time temperature rise rate sequence.

[0105] Step S530: Perform a time sequence comparison process on the temperature rise start collection time marker and the transition collection time marker that transitions from the saturated wet state to the continuous decline state. When the temperature rise start collection time marker is earlier than the transition collection time marker, take the temperature rise start collection time marker as the pre-correction marker of the transition collection time marker in the conductor surrounding dry tendency marker sequence.

[0106] If Ttemp<Tdry, it indicates that the surface temperature of the overhead conductor enters a rising or falling trend earlier than the relative humidity of the near-ground air, and the heating of the conductor may be a pre-factor leading to the drying of the near-ground air. Take Ttemp as the pre-correction marker of the transition collection time marker Tdry, and write it into the corresponding record of the conductor surrounding dry tendency marker sequence.

[0107] Step S540: Perform association and correspondence processing on the time-by-time temperature rise rate of the overhead conductor surface temperature within the continuous temperature rise period and the time-by-time decline rate of the ambient relative humidity in the overhead conductor passage within the continuous decline period, establish correspondence data pairs of the time-by-time temperature rise rate and the time-by-time decline rate, and add the correspondence data pairs as auxiliary criterion parameters to the conductor surrounding dry tendency marker sequence.

[0108] According to the collection time alignment principle, combine the time-by-time temperature rise rate ΔTk and the time-by-time decline rate ΔHk at the same collection time into a data pair (ΔTk, ΔHk). All data pairs are arranged in ascending order by collection time, and are added as auxiliary criterion parameters to the record of the corresponding co-located collection point in the conductor surrounding dry tendency marker sequence.

[0109] Step S610: After generating the vegetation combustible state evolution marker sequence, extract the frame-by-frame attenuation amplitude of the normalized vegetation index of each co-located collection point within the continuous attenuation period from the vegetation combustible state evolution marker sequence, arrange the frame-by-frame attenuation amplitude of each co-located collection point in the order of collection time, to form a frame-by-frame attenuation amplitude sequence.

[0110] Step S620: Perform attenuation mode decomposition processing on the frame-by-frame attenuation amplitude sequence of each co-located acquisition point, and classify the frame-by-frame attenuation amplitudes of multiple adjacent acquisition times in the frame-by-frame attenuation amplitude sequence into the same attenuation pulse event. When the frame-by-frame attenuation amplitudes of two adjacent acquisition times both exceed the preset pulse attenuation threshold, it is determined that the two adjacent acquisition times belong to the same attenuation pulse event. When the frame-by-frame attenuation amplitude of any one of the two adjacent acquisition times is lower than the preset pulse attenuation threshold, it is determined that the attenuation pulse event terminates.

[0111] The attenuation amplitude sequence is traversed frame by frame, and adjacent acquisition times where ΔNDVI continuously exceeds the preset pulse attenuation threshold Apulse are grouped into the same attenuation pulse event. The start acquisition time of the attenuation pulse event is the moment when it first exceeds Apulse, and the end acquisition time is the moment when it last exceeds Apulse.

[0112] Step S630: Extract the start and end acquisition times of each attenuation pulse event, calculate the acquisition time span between the start and end acquisition times as the duration of the attenuation pulse, and extract the peak value of the frame-by-frame attenuation amplitude of all acquisition times within the attenuation pulse event as the attenuation pulse intensity value.

[0113] The duration of the decay pulse is Dpulse = tend - tstart, and the intensity of the decay pulse is Ipulse = max(ΔNDVIk), where k ∈ the set of all acquisition time indices included in this decay pulse event.

[0114] Step S640: Combine the duration and intensity of the decay pulse for each decay pulse event to form the pulse characteristic parameters of that decay pulse event. Arrange the pulse characteristic parameters of all decay pulse events that occur sequentially during the continuous decay period of the vegetation canopy greenness characterization time series at the same co-located sampling point according to the order of occurrence of the decay pulse events to form the vegetation decay pulse event sequence for that co-located sampling point.

[0115] Step S650: Perform correlation analysis on the vegetation attenuation pulse event sequence of each co-located collection point and the fire hazard point marker type in the fire hazard alignment superposition relationship record set for that co-located collection point. Extract the vegetation attenuation pulse event sequence of the co-located collection points marked as fire hazard time-coupled points as fire hazard-related attenuation pulse event sequences, and extract the vegetation attenuation pulse event sequence of the co-located collection points marked as fire hazard time-far away points as non-fire hazard-related attenuation pulse event sequences.

[0116] Step S660: Compare the attenuation pulse duration and intensity value of each attenuation pulse event in the fire hazard-related attenuation pulse event sequence with the attenuation pulse duration and intensity value of each attenuation pulse event in the non-fire hazard-related attenuation pulse event sequence to extract the concentrated distribution intervals of the attenuation pulse duration and attenuation pulse intensity values ​​in the fire hazard-related attenuation pulse event sequence.

[0117] For the duration Dpulse of all attenuated pulses in the fire-related attenuated pulse event sequence, calculate its percentile range as the central distribution interval [DdurL, DdurU]. For the attenuated pulse intensity value Ipulse, calculate its percentile range as the central distribution interval [IintL, IintU].

[0118] Step S670: Combine the lower boundary value of the concentrated distribution interval of the attenuation pulse duration and the lower boundary value of the concentrated distribution interval of the attenuation pulse intensity value into the attenuation pulse fire hazard precursor characteristic parameter. When an attenuation pulse event occurs in the vegetation attenuation pulse event sequence of any co-located collection point to be monitored, where the attenuation pulse duration reaches the lower boundary value of the concentrated distribution interval of the attenuation pulse duration and the attenuation pulse intensity value simultaneously reaches the lower boundary value of the concentrated distribution interval of the attenuation pulse intensity value, a vegetation attenuation pulse fire hazard precursor marker is generated.

[0119] The characteristic parameters of attenuated pulse fire hazard precursors are (DdurL, IintL). When an attenuated pulse event at the monitoring point satisfies Dpulse≥DdurL and Ipulse≥IintL, a vegetation attenuated pulse fire hazard precursor marker is generated.

[0120] Step S680: Add the vegetation attenuation pulse fire hazard precursor marker as an auxiliary criterion to the lower boundary criterion for triggering the fire hazard warning in the power distribution corridor, and use the time node triggered by the vegetation attenuation pulse fire hazard precursor marker as the reference amount for the advance offset of the start time of the patrol period window planning, and update the patrol period window planning.

[0121] An auxiliary condition is added to the lower boundary criterion for triggering a fire hazard warning in the power distribution corridor: a fire hazard warning is triggered when the vegetation attenuation pulse fire precursor marker is true. The start time of the patrol period window is shifted forward by the time difference between the trigger time of the vegetation attenuation pulse fire precursor marker and the transition frame time marker Tndvi.

[0122] Step S710: After generating the drying tendency marker sequence around the conductor, extract the hourly decrease rate of the relative humidity in the overhead conductor channel of each co-located collection point during the continuous decrease period from the drying tendency marker sequence around the conductor. Arrange the hourly decrease rate of each co-located collection point in the order of collection time to form an hourly decrease rate sequence.

[0123] The time-by-time decrease rate sequence is extracted from the dryness tendency marking sequence around the conductor, according to the co-located collection point number. Each sequence is a one-dimensional array composed of the difference ΔH in the relative humidity of the overhead conductor channel at all collection times after the turning collection time mark Tdry at the co-located collection point, arranged in ascending order of collection time.

[0124] Step S720: Perform descent rate fluctuation rhythm detection processing on the time-by-time descent rate sequence of each co-located acquisition point, extract the fluctuation segments in the time-by-time descent rate sequence where the descent rate shows periodic high and low alternation, and mark the complete cycle of the descent rate in each fluctuation segment from local minimum to local maximum and then back to local minimum as a dry fluctuation rhythm cycle.

[0125] Local extrema were detected in the time-by-time descent rate sequence by comparing the magnitudes of three adjacent points along the sampling direction. If the descent rate at the midpoint was less than that of the two adjacent points, it was considered a local minimum; if the descent rate at the midpoint was greater than that of the two adjacent points, it was considered a local maximum. The interval between consecutive sampling points in the sequence, starting from a local minimum, passing through a local maximum, and then to the next local minimum, was marked as a dryness fluctuation rhythm period. This rhythmic fluctuation of first decreasing, then increasing, and then decreasing again reflects that the near-surface air drying process is not monotonous, but rather involves periodic alternations of intensified and alleviated dryness.

[0126] Step S730: Extract the duration of each drying fluctuation rhythm cycle and the fluctuation amplitude of the rate of decrease within the cycle. The duration of the cycle is the time span between the start and end times of the drying fluctuation rhythm cycle. The fluctuation amplitude of the rate of decrease within the cycle is the difference between the maximum and minimum values ​​of the rate of decrease within the cycle.

[0127] For each drying fluctuation rhythm cycle, the start and end times of the data collection are recorded, and the time difference between them is calculated as the cycle duration. The descent rate values ​​of all sampling points within the cycle are iterated, and the difference between the maximum and minimum values ​​is taken as the fluctuation amplitude.

[0128] Step S740: Combine the duration of each dryness fluctuation rhythm cycle and the fluctuation amplitude of the rate of decrease within the cycle to form the rhythm characteristic parameters of that dryness fluctuation rhythm cycle. Arrange the rhythm characteristic parameters of all dryness fluctuation rhythm cycles that appear sequentially during the continuous decrease period of the relative humidity time series in the overhead conductor channel at the same co-located sampling point according to the order of occurrence of the dryness fluctuation rhythm cycles to form the dryness fluctuation rhythm cycle sequence of that co-located sampling point.

[0129] The rhythmic characteristic parameters of each dryness fluctuation rhythm cycle are represented by a binary data structure, containing two fields: cycle duration and fluctuation amplitude. All dryness fluctuation rhythm cycles identified at the same co-located sampling point during the continuous decline period are arranged in chronological order of their respective starting sampling times, forming the dryness fluctuation rhythm cycle sequence for that co-located sampling point.

[0130] Step S750: Perform correlation analysis on the dryness fluctuation rhythm periodic sequence of each co-located collection point and the fire hazard point marker type in the fire hazard alignment superposition relationship record set. Extract the dryness fluctuation rhythm periodic sequence of the co-located collection points marked as fire hazard time-coupled points as fire hazard-related dryness fluctuation rhythm periodic sequences, and extract the dryness fluctuation rhythm periodic sequence of the co-located collection points marked as fire hazard time-far away points as non-fire hazard-related dryness fluctuation rhythm periodic sequences.

[0131] Based on the fire hazard point marker type of each co-located collection point in the fire hazard alignment and superposition relationship record set generated in step S140, all co-located collection points are divided into a fire hazard time-coupled point group and a fire hazard time-far away point group. The dryness fluctuation rhythm periodic sequences of the co-located collection points within each of the two groups are collected separately to form a fire hazard-related dryness fluctuation rhythm periodic sequence sample set and a non-fire hazard-related dryness fluctuation rhythm periodic sequence sample set. This grouping method is used to compare and analyze the differences in dryness fluctuation rhythm characteristics between channel segments where fire hazard coupling actually occurs and those where fire hazard coupling does not.

[0132] Step S760: Compare the duration of the cycle and the fluctuation amplitude of the rate of decline within the cycle of each dryness fluctuation rhythm cycle in the fire hazard-associated dryness fluctuation rhythm cycle sequence with the duration of the cycle and the fluctuation amplitude of the rate of decline within the cycle of each dryness fluctuation rhythm cycle in the non-fire hazard-associated dryness fluctuation rhythm cycle sequence. Extract the specific rhythmic feature parameter intervals in the fire hazard-associated dryness fluctuation rhythm cycle sequence that are different from those in the non-fire hazard-associated dryness fluctuation rhythm cycle sequence. The specific rhythmic feature parameter intervals include specific intervals for the duration of the cycle and specific intervals for the fluctuation amplitude of the rate of decline within the cycle.

[0133] The periodic duration and fluctuation amplitude distributions of the fire-related group and the non-fire-related group were statistically analyzed separately. The kernel density estimation algorithm was used to calculate the probability density functions of the two groups in the dimensions of periodic duration and fluctuation amplitude. In the dimension of periodic duration, the intervals where the probability density of the fire-related group was significantly higher than that of the non-fire-related group (i.e., the density ratio exceeded a preset density ratio threshold) were identified as specific intervals for periodic duration. The same method was used to extract specific intervals for fluctuation amplitude in the dimension of fluctuation amplitude. These specific intervals represent the unique rhythmic characteristics of the drying fluctuation rhythm under fire-related coupling scenarios.

[0134] Step S770: Combine the lower boundary value of the specific interval of the period duration and the lower boundary value of the specific interval of the fluctuation amplitude of the rate of descent within the period into the characteristic parameters of the dry fluctuation rhythm fire hazard precursor. When a dry fluctuation rhythm periodic sequence of any co-located acquisition point to be monitored has a dry fluctuation rhythm periodicity where the period duration falls within the specific interval of the period duration and the fluctuation amplitude of the rate of descent within the period falls within the specific interval of the fluctuation amplitude of the rate of descent within the period, a conductor dry rhythm fire hazard precursor marker is generated.

[0135] The characteristic parameters of the dryness fluctuation rhythm fire hazard precursor are binary, including the lower boundary value of the period duration-specific interval and the lower boundary value of the fluctuation amplitude-specific interval. The dryness fluctuation rhythm periodic sequence of the co-located data collection points to be monitored is scanned in real time. When the period duration of a certain dryness fluctuation rhythm period exceeds the lower boundary value of both the period duration and the fluctuation amplitude exceeds the lower boundary value of the fluctuation amplitude, the conductor dryness rhythm fire hazard precursor marker is triggered.

[0136] Step S780: Add the conductor dry rhythm fire hazard precursor marker as an auxiliary criterion to the lower boundary criterion for triggering fire hazard warning in the power distribution corridor, and use the time node triggered by the conductor dry rhythm fire hazard precursor marker as the post-offset reference amount of the end time of the patrol period window planning, and update the patrol period window planning.

[0137] An auxiliary criterion is added to the lower boundary criterion for triggering fire hazard warnings in power distribution corridors: when the conductor drying rhythm fire hazard precursor marker is true, a fire hazard warning is triggered even if the frame-by-frame attenuation amplitude and time-by-time decrease rate have not yet reached the lower boundary criterion. The termination time of the patrol period window is shifted backward by the time difference between the trigger time of the conductor drying rhythm fire hazard precursor marker and the turning point acquisition time marker Tdry, to ensure that the patrol period covers the abnormal cycle of the drying fluctuation rhythm.

[0138] Step S810: After obtaining the vegetation combustible state evolution marker sequence and the conductor perimeter drying tendency marker sequence, extract the transition frame time marker of the normalized vegetation index of each co-located collection point from the vegetation combustible state evolution marker sequence, which indicates the transition from a stable fluctuating state to a continuously declining state; and extract the transition collection time marker of each co-located collection point from a saturated moist state to a continuously declining state from the conductor perimeter drying tendency marker sequence.

[0139] Step S820: Perform a chronological sequence comparison process on the turning frame time marker and the turning collection time marker of the same co-located collection point, mark a co-located collection point where the turning frame time marker is earlier than the turning collection time marker as a vegetation-prior point, mark a co-located collection point where the turning frame time marker is later than the turning collection time marker as a drying-prior point, and mark a co-located collection point where the absolute value of the time difference between the turning frame time marker and the turning collection time marker is smaller than a preset synchronization determination threshold as a synchronous trigger point.

[0140] For each co-located collection point, read Tndvi and Tdry. If Tndvi < Tdry and the absolute value of the difference exceeds the preset synchronization determination threshold, the point is marked as a vegetation-prior point, indicating that vegetation greenness attenuation occurs before near-surface air drying. If Tdry < Tndvi and the absolute value of the difference exceeds the preset synchronization determination threshold, the point is marked as a drying-prior point, indicating that near-surface air drying occurs before vegetation greenness attenuation. If the absolute value of the difference does not exceed the preset synchronization determination threshold, the point is marked as a synchronous trigger point, indicating that the two events occur almost simultaneously.

[0141] Step S830: Extract the absolute value of the time difference between the turning frame time marker and the turning collection time marker of all vegetation-prior points, calculate the distribution center value of the absolute time difference values of all vegetation-prior points as the vegetation-prior time difference representative value, extract the absolute value of the time difference between the turning collection time marker and the turning frame time marker of all drying-prior points, and calculate the distribution center value of the absolute time difference values of all drying-prior points as the drying-prior time difference representative value.

[0142] The vegetation-prior time difference representative value is taken as the median of the difference between Tdry and Tndvi of all vegetation-prior points. The drying-prior time difference representative value is taken as the median of the difference between Tndvi and Tdry of all drying-prior points. This representative value represents the typical lag time between the two turning events in the vegetation-prior or drying-prior scenario.

[0143] Step S840: Construct a vegetation-prior trigger mode criterion according to the vegetation-prior time difference representative value, wherein the vegetation-prior trigger mode criterion is used to perform an early warning of drying turning for the environmental relative humidity in the overhead line channel of the co-located collection point after the time span corresponding to the vegetation-prior time difference representative value elapses from the occurrence of the turning frame time marker when the normalized difference vegetation index of any co-located collection point to be monitored changes from a stable fluctuation state to a continuous attenuation state.

[0144] The criteria for vegetation leader triggering mode stipulate that when a turning frame time marker is detected at a co-located collection point, indicating that the vegetation has entered a state of continuous decay, a drying turning point warning will be issued in advance after the time span corresponding to the vegetation leader time difference value, even if the relative humidity of the environment in the overhead conductor channel at that point has not yet shown a drying turning point. This indicates that the point may be about to enter a drying state driven by vegetation decay.

[0145] Step S850: Construct a dry leader triggering mode criterion based on the dry leader time difference representative value. The dry leader triggering mode criterion is used to provide an early warning of the flammability transition of the vegetation canopy greenness characterization of any co-located sampling point after the time span corresponding to the dry leader time difference representative value appears when the transition sampling time mark of any co-located sampling point from a saturated humid state to a continuously decreasing state occurs.

[0146] The criteria for the dry leader trigger mode stipulate that when a turning point in the near-ground air is detected at a co-located sampling point, indicating that the air has entered a state of continuous dryness, a flammability transition warning is issued in advance after the time span corresponding to the dry leader time difference value, even if the vegetation greenness at that point has not yet shown a decline transition. This warning indicates that the point may be about to enter a flammable state under the stress of a dry environment.

[0147] Step S860: Extract the spatial coordinates of all synchronous triggering points, calculate the spatial clustering density of synchronous triggering points along the overhead power distribution channel, and mark the section where synchronous triggering points are continuously clustered as a fire hazard synchronous triggering sensitive corridor section when the spatial clustering density exceeds the preset spatial clustering density threshold.

[0148] The spatial coordinates of the synchronously triggered points are projected one-dimensionally along the power distribution line, and the density of synchronously triggered points at each location is calculated using a sliding window density estimation algorithm. Continuous sections with densities exceeding a preset spatial aggregation density threshold are marked as fire-sensitive synchronous triggering corridors, indicating that the dry vegetation and flammable state occur almost simultaneously in these corridors, resulting in the most rapid fire-coupled response.

[0149] Step S870: Add the vegetation-led triggering mode criterion, the dryness-led triggering mode criterion, and the fire-triggered sensitive corridor segment marker to the fire warning triggering criterion data of the power distribution corridor, so that the fire warning triggering criterion data of the power distribution corridor includes classified warning criteria under different triggering time sequence modes.

[0150] The updated fire hazard warning triggering criteria data for power distribution corridors includes three sub-criteria modules: vegetation-led triggering mode criteria, dryness-led triggering mode criteria, and fire-synchronous triggering sensitive corridor segment markers. During fire hazard monitoring, the corresponding sub-criteria is invoked for differentiated warnings based on the triggering sequence type of the current co-located data collection point.

[0151] Step S910: After generating the fire hazard alignment and superposition relationship record set, extract the fire hazard point marker type of each co-located collection point from the fire hazard alignment and superposition relationship record set. Extract the spatial coordinates of the co-located collection points with the fire hazard point marker type of fire hazard time coupling point into a fire hazard coupling point coordinate set. Extract the spatial coordinates of the co-located collection points with the fire hazard point marker type of fire hazard time adjacent point into a fire hazard adjacent point coordinate set. Extract the spatial coordinates of the co-located collection points with the fire hazard point marker type of fire hazard time far away point into a fire hazard far away point coordinate set.

[0152] Step S920: Perform spatial continuous region growth processing on the set of fire hazard coupling point coordinates, and group fire hazard coupling points with spatial distances less than the preset spatial proximity distance threshold into the same fire hazard coupling continuous region, generating multiple fire hazard coupling continuous region blocks, and calculating the coverage area of ​​each fire hazard coupling continuous region block and the number of fire hazard coupling points in the region.

[0153] A density-based spatial clustering algorithm is used for region growing. Each fire hazard coupling point is used as a seed point, and the algorithm expands outwards in two directions: along the power line and perpendicular to the power line. Other fire hazard coupling points with a spatial distance less than a preset spatial proximity threshold are included in the same continuous region. After region growing is completed, each fire hazard coupling continuous region block is assigned a unique identifier, its coverage area is the area enclosed by the region boundary, and the number of points is the total number of fire hazard coupling points contained in that region.

[0154] Step S930: Perform spatial continuous region growing processing on the set of coordinates of fire hazard adjacent points, and group fire hazard adjacent points whose spatial distance is less than the preset spatial adjacent distance threshold into the same fire hazard adjacent continuous region, generating multiple fire hazard adjacent continuous region blocks, and calculating the coverage area of ​​each fire hazard adjacent continuous region block and the number of fire hazard adjacent points in the region.

[0155] Using the same region growth method as in step S920, generate fire hazard adjacent continuous region blocks and calculate the coverage area and number of points.

[0156] Step S940: Perform spatial continuous region growth processing on the set of fire hazard distance coordinates, group fire hazard distances with spatial distances less than a preset spatial proximity threshold into the same fire hazard distance continuous region, generate multiple fire hazard distance continuous region blocks, and calculate the coverage area of ​​each fire hazard distance continuous region block and the number of fire hazard distances within the region.

[0157] Using the same region growth method as in step S920, generate fire hazard-free continuous region blocks and calculate the coverage area and number of points.

[0158] Step S950: Perform spatial adjacency determination on the spatial boundary of each fire hazard coupled continuous area block and the spatial boundary of the fire hazard adjacent continuous area block. When there is spatial boundary overlap between the fire hazard coupled continuous area block and the fire hazard adjacent continuous area block or the spatial boundary distance is less than the preset area adjacency determination distance, merge the fire hazard coupled continuous area block and the fire hazard adjacent continuous area block into the same fire hazard composite risk area.

[0159] The spatial boundaries of fire hazard coupled continuous area blocks and fire hazard adjacent continuous area blocks are checked one by one. If the boundaries of the two blocks intersect, or the shortest distance between the two boundaries is less than the preset area adjacency determination distance, the two blocks are determined to be spatially adjacent and merged into a single fire hazard composite risk area. The spatial boundary of the merged area is the convex hull of the union of the two boundaries.

[0160] Step S960: The sum of the coverage area of ​​the fire-coupled continuous area block and the coverage area of ​​the fire-adjacent continuous area block contained in each fire risk composite risk area is taken as the comprehensive risk coverage area of ​​the fire risk composite risk area. The comprehensive risk coverage area of ​​the fire risk composite risk area is compared with the preset graded area threshold. The fire risk composite risk areas with a comprehensive risk coverage area greater than the preset graded area threshold are marked as key fire risk prevention and control areas.

[0161] The larger the comprehensive risk coverage area of ​​a fire hazard complex risk zone, the denser the fire hazard coupling and adjacent fire hazard points within the area, and the greater the probability of fire and the wider the impact. Areas with a comprehensive risk coverage area exceeding a preset classification area threshold are marked as key fire hazard prevention and control areas.

[0162] Step S970: Extract the boundary coordinates of the key fire hazard prevention and control area, expand the boundary coordinates of the key fire hazard prevention and control area outward by a preset buffer distance along the outer normal direction to obtain the firebreak widening area, and add the firebreak widening area to the firebreak opening location guide.

[0163] For the boundary polygon of the key fire risk prevention and control area, each side is translated outward by a preset buffer distance along the outward normal direction. The intersection of the translated side and the adjacent side constitutes the widened area boundary. The widened area is added to the firebreak opening location guide generated in step S156 as the firebreak opening location guide and marked as a priority opening area.

[0164] Step S1010: After generating the fire hazard warning trigger criterion data for the power distribution corridor, obtain the frame-by-frame decay amplitude of the normalized vegetation index at each co-located collection point along the power distribution overhead channel during the continuous decay period, and the hour-by-hour decrease rate of the relative humidity in the overhead conductor channel during the continuous decrease period.

[0165] Step S1020: Input the frame-by-frame attenuation amplitude and time-by-time descent rate of each co-located acquisition point into the bivariate fire risk situation evolution state space. The bivariate fire risk situation evolution state space uses the frame-by-frame attenuation amplitude as the first state dimension coordinate axis and the time-by-time descent rate as the second state dimension coordinate axis. After performing dimensionless standardization on the data of the two coordinate axes, map them to a unified scale state space. Use the normalized relative value of the frame-by-frame attenuation amplitude as the first axis and the normalized relative value of the time-by-time descent rate as the second axis, so that the numerical contributions of the two axes in the state space are commensurable.

[0166] The normalization method for the frame-by-frame attenuation amplitude is to divide the frame-by-frame attenuation amplitude value of each co-located acquisition point by the upper bound reference value Vupper for vegetation yellowing and spreading rate, thus mapping it to the interval between 0 and 1. The normalization method for the time-by-time decline rate is to divide the time-by-time decline rate value by the upper bound reference value Dupper for environmental drying rate, thus mapping it to the interval between 0 and 1. The normalization process eliminates the dimensional difference between the two dimensions of vegetation attenuation and drying rate, making them comparable in state space.

[0167] Step S1030: Map the dimensionless standardized frame-by-frame attenuation amplitude and time-by-time descent rate values ​​of the same co-located acquisition point at multiple consecutive acquisition times to the bivariate fire risk situation evolution state space to form a fire risk situation state point sequence for the co-located acquisition point. Connect adjacent state points in the fire risk situation state point sequence with directed line segments according to the order of acquisition time to generate the fire risk situation evolution trajectory.

[0168] Using the normalized frame-by-frame attenuation amplitude at each acquisition moment as the x-axis and the normalized time-by-time descent rate as the y-axis, a state point is determined in the bivariate fire risk situation evolution state space. Adjacent state points are connected sequentially by directed line segments according to the acquisition time sequence, forming a continuous directed trajectory from the initial moment to the current moment, which reflects the evolution path of the fire risk situation at the co-located acquisition point over time.

[0169] Step S1040: Extract the trajectory morphology features of the fire hazard situation evolution trajectory for each co-located collection point, and extract the migration direction angle and total migration path length of the fire hazard situation evolution trajectory from the origin of the state space to the direction away from the origin. The migration direction angle is the angle between the line connecting the start and end points of the fire hazard situation evolution trajectory and the first state dimension coordinate axis. The total migration path length is the sum of the lengths of the directed line segments on the fire hazard situation evolution trajectory.

[0170] The direction of migration reflects whether the main trend of fire hazard evolution is driven by vegetation decline or by drying rate. The total length of the migration path reflects the degree of fluctuation and detour experienced by the fire hazard during its evolution.

[0171] Step S1050: Extract the local area with the highest density of state points in the fire hazard situation evolution trajectory of each co-located collection point as the fire hazard situation hovering area, calculate the coordinate value of the center point of the fire hazard situation hovering area in the bivariate fire hazard situation evolution state space, take the first state dimension coordinate value of the center point of the fire hazard situation hovering area as the vegetation attenuation hovering level value, and take the second state dimension coordinate value of the center point of the fire hazard situation hovering area as the drying rate hovering level value.

[0172] The spatial distribution density of state points on the fire hazard evolution trajectory is calculated using a kernel density estimation algorithm. The region with the highest density is the fire hazard lingering region, indicating that the fire hazard situation remains in that region for the longest time. The x-axis of the center point of the lingering region represents the vegetation decay lingering level value, and the y-axis represents the drying rate lingering level value.

[0173] Step S1060: Combine the migration direction angle, total migration path length, vegetation attenuation level, and drying rate attenuation level of each co-located collection point into a fire risk situation evolution feature vector for that co-located collection point.

[0174] Step S1070: Extract the fire hazard situation evolution feature vectors of co-located collection points marked as fire hazard time coupling points in the fire hazard alignment and superposition relationship record set as the coupling class feature vector set; extract the fire hazard situation evolution feature vectors of co-located collection points marked as fire hazard time far away points in the fire hazard alignment and superposition relationship record set as the far away class feature vector set.

[0175] Step S1080: Solve the feature space separation surface for the coupled feature vector set and the remote feature vector set to obtain the migration direction angle boundary value and the migration path total length boundary value that distinguishes between coupled feature vectors and remote feature vectors. Use the migration direction angle boundary value and the migration path total length boundary value as the fire hazard situation evolution trajectory early warning boundary parameters.

[0176] Classification algorithms such as Support Vector Machine (SVM) or Linear Discriminant Analysis (LDA) are used to solve for the classification hyperplane in the feature space that best distinguishes between coupled and distant class samples. The intersection of the classification hyperplane with the migration direction angle axis is extracted as the migration direction angle boundary value, and the intersection of the classification hyperplane with the total migration path length axis is extracted as the total migration path length boundary value.

[0177] Step S1090: When the migration direction angle of the fire hazard situation evolution trajectory of any co-located data collection point to be monitored exceeds the migration direction angle boundary value and the total migration path length exceeds the total migration path length boundary value, a fire hazard situation trajectory boundary crossing warning mark is generated and the fire hazard situation trajectory boundary crossing warning mark is added to the fire hazard warning triggering criterion data of the power distribution corridor.

[0178] When the migration direction angle of the fire hazard situation evolution trajectory is greater than the angle boundary value and the total length of the migration path is greater than the path boundary value, it indicates that the fire hazard situation evolution at this point has entered the distribution area of ​​coupling characteristics, with a high probability of fire hazard coupling. This triggers the fire hazard situation trajectory boundary crossing warning mark, which is added as a new criterion to the fire hazard warning triggering criterion data of the power distribution corridor, and is used to predict the fire hazard coupling risk in advance from the perspective of evolution trajectory.

[0179] like Figure 2 As shown, in steps S110 to S130, a co-located acquisition point array is first deployed along the overhead power distribution channel, with each point equipped with a multispectral image acquisition device and an environmental temperature and humidity sensor. Figure 2 The first layer displays two types of data synchronously acquired through the aforementioned equipment: a multispectral image frame sequence of the vegetation canopy (including the red band R and the near-infrared band NIR) and a temperature and humidity sensing sequence of the environment inside the overhead conductor channel. The two types of data are combined into a synchronously acquired data package.

[0180] When performing step S120, as Figure 2 As shown in the second layer, the vegetation canopy multispectral image frame sequence is extracted from the synchronously acquired data packet, the Normalized Difference Vegetation Index (NDVI) is calculated, and an NDVI time series is generated. For example... Figure 2 As shown in the line chart, this time series exhibits a stable fluctuation state with small fluctuations in the early stage, followed by a continuous downward decay state in the later stage. Through stationarity transition detection processing, the following characteristics are extracted: Figure 2 The red inverted triangle in the center indicates the transition time marker, marking the point at which the vegetation canopy greenness shifts from stable fluctuations to continuous decline. Subsequently, the difference Δ between the NDVI values ​​at adjacent acquisition times is calculated after this transition time marker, forming... Figure 2 The decreasing histogram sequence is used as the frame-by-frame decay amplitude. The aforementioned transition frame time markers and frame-by-frame decay amplitudes together constitute the vegetation combustible state evolution marker sequence.

[0181] When performing step S130, as Figure 2 As shown in the third layer, the environmental temperature and humidity sensor sequence is extracted from the synchronously acquired data packets to generate a relative humidity time series. Figure 2 As shown in the line graph, the relative humidity time series exhibits a high and flat saturated humid state in the first half, followed by a continuous downward trend in the second half. By identifying the saturated humid state and tracking the drying trend, the following parameters were extracted: Figure 2 The blue inverted triangle indicates the turning point in relative humidity, marking the transition from saturated humidity to a continuous decrease. Subsequently, the difference in relative humidity decrease between adjacent moments is calculated, forming... Figure 2The decreasing histogram sequence represents the hourly rate of descent. These inflection point acquisition time markers and the hourly rate of descent together constitute the drying tendency marker sequence around the conductor.

[0182] like Figure 3 As shown, in steps S140 to S150, the two types of marker sequences are spatiotemporally fused and boundary extracted. Figure 3 The left side illustrates the fire hazard spatiotemporal alignment and overlay processing in step S140. A three-dimensional grid matrix is ​​constructed with spatial point numbers (1 to N) as the X-axis, time axis (t0 to tn) as the Y-axis, and overlay state value as the Z-axis. After inputting the vegetation combustible state evolution marker sequence (red cards) and the conductor perimeter drying tendency marker sequence (blue cards) into this matrix, the matrix cells are classified and filled according to the time interval between the two types of markers at the same spatial point: those with an interval less than the preset coupling judgment time window are filled with a red diagonal texture (fire hazard time coupled points), those with an interval between the coupling and adjacent judgment time windows are filled with an orange grid texture (fire hazard time adjacent points), and those with an interval greater than the adjacent judgment time window are filled with a gray dotted texture (fire hazard time far away points). This outputs the set of fire hazard alignment and overlay relationship records.

[0183] Figure 3 The right side shows the fire hazard coupling trigger boundary extraction in step S150. A bivariate scatter plot is constructed with the frame-by-frame attenuation magnitude as the horizontal axis and the time-by-time descent rate as the vertical axis. The red scatter points in the plot are the fire hazard time coupling points, which cluster to form a red dashed box area. Four key reference values ​​are extracted from the boundaries of this area: the upper boundary reference value of the vegetation withering and spreading rate, the lower boundary reference value of the vegetation withering and spreading rate, the right boundary reference value of the environmental drying rate, and the left boundary reference value of the environmental drying rate. Using these boundary values, the internal area of ​​the scatter plot is divided into a light red area (lower boundary criterion, triggering fire hazard attention warning) and a dark red area (upper boundary criterion, triggering fire hazard emergency alarm), finally obtaining the fire hazard warning trigger criterion data for the power distribution corridor.

[0184] like Figure 4 As shown, in step S150, a fire risk management strategy is further generated based on the fire risk alignment and superposition relationship record set. Figure 4 The main image illustrates the process of generating the location guidance for firebreaks. The spatial coordinates of fire hazard time-coupled points are extracted, and their continuous, densely packed sections within the power distribution corridor are identified. Figure 4 (Mid-red semi-transparent covering layer). Extend the starting coordinates of this continuous dense section outward in the opposite direction of the power distribution line, and extend the ending coordinates outward in the positive direction (as shown by the yellow dashed arrow). The area outside the corridor boundary covered by the extension is the location guide for the fire isolation zone (yellow diagonal filled area).

[0185] Figure 4 The timeline chart on the right illustrates the generation process of the patrol time window plan. The time nodes when vegetation enters a combustible state (red blocks) and the time nodes when the perimeter of the firebreak enters a dry state (blue blocks) are extracted for each fire hazard time coupling point. The overlap between these two time nodes is calculated. This overlap is the highlighted block, with its left end corresponding to the patrol start time and its right end corresponding to the patrol end time, forming the patrol time window plan. The aforementioned firebreak location guidance and the patrol time window plan together constitute... Figure 4 The fire risk management strategy is shown at the bottom.

[0186] Following step S1090, further, referring to Figure 5 The diagram shows a comparison of the spatial trajectories of the evolution of the bivariate fire risk situation. Figure 5 The horizontal axis represents the frame-by-frame attenuation magnitude (unit: NDVI attenuation / frame), and the vertical axis represents the time-by-time decrease rate (unit: % / time). The figure shows the evolution trajectory of three typical fire hazard location categories. Among them, the red solid arrow trajectory represents the fire risk time coupling point A001. Its trajectory starts from the initial state point (0.03, 0.8%), passes through state points such as (0.05, 1.2%) and (0.07, 1.9%), and finally reaches (0.18, 4.5%). The overall migration direction angle is 62° and the total migration path length is 5.2, showing a clear trend of migration towards the high attenuation and high descent rate area. The black dashed arrow trajectory represents the fire risk time far away point B012. Its trajectory slowly extends from (0.01, 0.3%) to (0.04, 1.0%). The migration direction angle is only 15° and the total migration path length is only 0.8, always hovering in the low-risk area. The green dotted arrow trajectory represents the fire risk time adjacent point C008. Its trajectory extends from (0.02, 0.5%) to (0.09, 1.8%). The migration direction angle is 38° and the total migration path length is 2.1, which is between the former two. The diagram also marks a red dashed line indicating the migration direction angle boundary of 45° and a black vertical dashed line indicating the total migration path length boundary of 2.5, dividing the state space into four quadrants. The upper right quadrant is covered in light red and marked as the "fire hazard trajectory boundary crossing warning area." Simulation results show that, using these two boundary values ​​as criteria, the accuracy rate for identifying fire hazard time-coupled points reaches 92.3%, and the exclusion rate for fire hazard time-far from points reaches 95.1%.

[0187] Furthermore, referring to Figure 6 The diagram shown is a joint analysis of vegetation attenuation pulse events and aridity fluctuation rhythms. Figure 6The graph consists of an upper panel with a bar chart overlaid with a line chart and a lower panel with a heatmap. In the upper panel, the horizontal axis represents the acquisition time t1 to t31, the left vertical axis represents the frame-by-frame attenuation amplitude, and the right vertical axis marks pulse events. The graph uses blue bars to represent the frame-by-frame attenuation amplitude at each time point, and highlights attenuation pulse events in red for specific time periods, such as "Pulse Event 1" from t7 to t10 and "Pulse Event 2" from t16 to t22. Analysis shows that the duration of fire hazard-related attenuation pulse events is concentrated between 3 and 7 acquisition times, with a lower boundary value of 3; the pulse intensity values ​​are concentrated between 0.12 and 0.21, with a lower boundary value of 0.12. The lower panel is a heatmap, with the horizontal axis representing the acquisition time and the vertical axis representing co-located acquisition points P01 to P20. The color changes from dark blue to red, reflecting the rate of attenuation at each time point. The figure shows that at fire hazard time-coupled locations P01 to P05, three alternating orange-red bands appeared between t9 and t15, with a corresponding cycle duration of approximately 3 sampling times and a fluctuation amplitude of 1.8 percentage points per moment. Between t16 and t22, two alternating red-orange bands appeared, with a cycle duration of approximately 3 sampling times and a fluctuation amplitude of 2.1 percentage points per moment. In contrast, the rhythmic characteristics of fire hazard time-coupled locations are relatively flat. By extracting these specific rhythmic characteristic parameter ranges, the lower boundary values ​​of the cycle duration and fluctuation amplitude can be combined into dry fluctuation rhythm fire hazard precursor characteristic parameters, which can be used to assist in triggering fire hazard attention warnings.

[0188] Furthermore, referring to Figure 7 The diagram shown is a comparison of spatial clustering and response strategies for fire hazard complex risk areas in power distribution corridors. Figure 7 The left side shows the spatial clustering diagram of the fire hazard composite risk area in the power distribution corridor. The horizontal axis represents the corridor distance (km), and the vertical axis represents the spatial distribution of risk points. Black dots represent co-located data collection points, with solid black circles representing fire hazard time-coupled points, gray solid circles representing fire hazard time-adjacent points, and dark gray solid circles representing fire hazard time-far away points. Through spatial continuous region growing processing, fire hazard coupled points with a spatial distance less than a preset proximity threshold are grouped into the same continuous region. The diagram shows three main fire hazard coupled continuous region blocks (A, B, C), circled by red dashed ellipses. Simultaneously, the fire hazard coupled continuous region blocks are spatially adjacent to fire hazard adjacent continuous region blocks and merged to generate fire hazard composite risk areas, such as the light red dashed circle marking the area surrounding region A and region B. Figure 7The right side shows a comparison of four key indicators before and after adopting this method. The bar chart compares the indicator values ​​before planning (fixed-cycle patrols) and after planning (the time window of this method), showing that the average early warning response time decreased by 75% (from 48 to 12), the effective coverage area of ​​a single patrol increased by 167% (from 35 to 92), the false alarm rate decreased by 64% (from 42 to 15), and the resource input efficiency increased by 167% (from 2.1 to 5.6). The table below further summarizes the comparison results of these four key indicators, verifying the significant advantages of this method in improving response speed, expanding coverage, reducing false alarm rate, and improving resource efficiency.

[0189] In an exemplary embodiment, a smart assessment system for fire risk of power distribution lines in forest and grassland areas is provided. This system can be a terminal, server, etc., and its internal structure includes a processor, memory, input / output interface, communication interface, display unit, and input device. The processor, memory, and input / output interface are connected via a system bus, and the communication interface, display unit, and input device are also connected to the system bus via the input / output interface. The processor provides computing and control capabilities. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and computer programs. The internal memory provides an environment for the operation of the operating system and computer programs in the non-volatile storage medium. The input / output interface is used for exchanging information between the processor and external devices. The communication interface is used for wired or wireless communication with external terminals; wireless communication can be achieved through Wi-Fi, mobile cellular networks, near-field communication, or other technologies. When the computer program is executed by the processor, it implements a smart assessment method for fire risk of power distribution lines in forest and grassland areas. The display unit is used to generate a visually visible image and can be a display screen, projection device, or virtual reality imaging device. The display screen can be an LCD screen or an e-ink screen. The input device can be a touch layer covering the display screen, or a button, trackball, or touchpad set on the casing of the intelligent assessment system for fire risk of power distribution lines facing forest and grassland areas, or an external keyboard, touchpad, or mouse, etc.

[0190] It should be noted that, in order to simplify the description of the present invention and thus help to understand one or more embodiments of the invention, multiple features may sometimes be grouped into one embodiment, drawing or description thereof in the foregoing description of the embodiments of the present invention.

Claims

1. A method for intelligent assessment of fire risk of power distribution lines in forest and grassland areas, characterized in that, The method includes: Acquire synchronous data packets of the forest and grassland area to be evaluated within the overhead power line channel. The synchronous data packets include a sequence of multispectral images of vegetation canopy continuously acquired along the power line and a sequence of environmental temperature and humidity sensing data within the overhead power line channel. The synchronously acquired data packets are subjected to vegetation canopy water stress evolution analysis processing to obtain vegetation combustible state evolution marker sequence. The vegetation combustible state evolution marker sequence is used to record the transition frame time marker of the normalized vegetation index from a stable fluctuation state to a continuous decay state in the vegetation canopy multispectral image frame and the frame-by-frame decay amplitude of the normalized vegetation index during the continuous decay period. The synchronously acquired data packets are processed to track the drying trend of the near-ground environment of the overhead conductor, resulting in a drying tendency marker sequence around the conductor. The vegetation combustible state evolution marker sequence and the drying tendency marker sequence around the conductor are input into the fire hazard spatiotemporal alignment superposition process to generate a fire hazard alignment superposition relationship record set. The fire hazard alignment and superposition relationship record set is processed by fire hazard coupling trigger boundary extraction to obtain fire hazard warning trigger criterion data for power distribution corridors. Based on the fire hazard warning trigger criterion data for power distribution corridors, a fire risk management strategy including guidance on the location of fire isolation zones and patrol time window planning is generated.

2. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The acquisition of synchronously collected data packets of the forest and grassland area to be evaluated within the overhead power line corridor includes: Along the overhead power line corridor, a co-located data acquisition point array is set up at a preset spatial interval. Each co-located data acquisition point is equipped with a multispectral image acquisition device and an environmental temperature and humidity sensor. The multispectral image acquisition device at each co-located acquisition point is driven to take pictures of the vegetation canopy in the forest and grassland area below the overhead power distribution channel from a top-down direction, and to obtain the vegetation canopy multispectral image frame corresponding to the co-located acquisition point. The vegetation canopy multispectral image frame includes red band reflectance value and near-infrared band reflectance value. The vegetation canopy multispectral image frames acquired at each co-located acquisition point at multiple consecutive acquisition times are arranged in chronological order of acquisition time to form the vegetation canopy multispectral image frame fragment corresponding to that co-located acquisition point. The environmental temperature and humidity sensing device in each co-located collection point is driven to sense and measure the relative humidity of the air environment in the near-ground space within the overhead conductor channel, and obtain the relative humidity value of the overhead conductor channel corresponding to the co-located collection point. The relative humidity values ​​of the overhead conductor channel obtained by each co-located collection point at multiple consecutive collection times are arranged in chronological order of collection time to form a relative humidity segment of the overhead conductor channel corresponding to the co-located collection point. Each multispectral image frame within the vegetation canopy multispectral image frame segment corresponding to the same co-located acquisition point is associated with the relative humidity value of the environment in the overhead conductor channel at the same acquisition time, generating a corresponding record of the image frames and environmental humidity of the co-located acquisition point. The corresponding records of the image frames and environmental humidity generated by all co-located acquisition points are arranged according to the spatial order of the power distribution line, and combined into a sequence of vegetation canopy multispectral image frames and a sequence of environmental temperature and humidity sensing in the overhead conductor channel. The multispectral image frame sequence of the vegetation canopy and the environmental temperature and humidity sensing sequence within the overhead conductor channel were packaged into a synchronous acquisition data package.

3. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The synchronously acquired data packets are subjected to vegetation canopy water stress evolution analysis to obtain vegetation combustible state evolution marker sequences, including: Extract the vegetation canopy multispectral image frame fragments corresponding to each co-located acquisition point from the synchronous acquisition data packet. Extract the red band reflectance value and near-infrared band reflectance value for each multispectral image frame in the vegetation canopy multispectral image frame fragment. Calculate the normalized vegetation index based on the red band reflectance value and near-infrared band reflectance value of each multispectral image frame. Use the normalized vegetation index as the vegetation canopy greenness characterization of that multispectral image frame. The vegetation canopy greenness characteristics of each multispectral image frame in the same co-located collection point are arranged in chronological order of collection time to generate a time series of vegetation canopy greenness characteristics. The time series of vegetation canopy greenness characteristics is processed by stationarity transition detection to extract the transition point collection time when the vegetation canopy greenness characteristics change from a stable fluctuation state with fluctuation amplitude less than a preset stationarity threshold to a continuous monotonically decreasing state. The transition point collection time is recorded as the transition frame time marker when the normalized vegetation index changes from a stable fluctuation state to a continuous decay state. Starting from the transition frame time mark, extract the vegetation canopy greenness characterization value for each subsequent acquisition time in the vegetation canopy greenness characterization value time series along the acquisition time direction. Calculate the difference between the vegetation canopy greenness characterization values ​​of two adjacent acquisition times and record the difference as the frame-by-frame attenuation value. Arrange the frame-by-frame attenuation values ​​corresponding to each acquisition time in the continuous period after the transition frame time mark according to the order of acquisition time to form a frame-by-frame attenuation value sequence. Record the frame-by-frame attenuation value sequence as the frame-by-frame attenuation amplitude of the normalized vegetation index during the continuous attenuation period. The transition frame time marker of the normalized vegetation index (NDI) from a stable fluctuation state to a continuously decaying state corresponding to each co-located collection point, and the frame-by-frame decay amplitude of the NDI during the continuously decaying period are combined to form the vegetation combustible state evolution marker record for that co-located collection point. The vegetation combustible state evolution marker records of all co-located collection points are summarized according to spatial location to generate a vegetation combustible state evolution marker sequence.

4. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The synchronously acquired data packets are processed to track the drying trend of the near-ground environment of the overhead conductor, resulting in a drying tendency marker sequence around the conductor, including: Extract the relative humidity segment of the overhead conductor channel corresponding to each co-located acquisition point from the synchronous acquisition data packet, and arrange the relative humidity values ​​of the overhead conductor channel in the relative humidity segment of the overhead conductor channel in the order of acquisition time to form the time series of relative humidity of the overhead conductor channel. Saturated humidity state identification processing is performed on the time series of relative humidity in the overhead conductor channel. Continuous segments in which the relative humidity value in the overhead conductor channel is continuously higher than the preset saturated humidity threshold are extracted, and the last collection time in the continuous segment is recorded as the end time of the saturated humidity state. Starting from the next acquisition time after the end of the saturated humid state, the relative humidity time series of the environment inside the overhead conductor channel is tracked backward along the acquisition time direction. The acquisition time when the relative humidity value inside the overhead conductor channel first falls below the preset drying trigger threshold and the relative humidity value inside the overhead conductor channel remains below the preset drying trigger threshold for multiple consecutive acquisition times is recorded as the transition acquisition time mark from the saturated humid state to the continuously decreasing state. Starting from the turning point acquisition time mark, extract the relative humidity value of the overhead conductor channel at each subsequent acquisition time in the time series of the overhead conductor channel along the acquisition time direction, and calculate the decrease difference between the relative humidity value of the overhead conductor channel at each acquisition time and the relative humidity value of the overhead conductor channel at the previous acquisition time. After marking the turning point acquisition time, the decrease difference corresponding to each acquisition time in the continuous period is arranged in the order of acquisition time to form a time-by-time decrease difference sequence. The time-by-time decrease difference sequence is recorded as the time-by-time decrease rate of the relative humidity of the environment in the overhead conductor channel during the continuous decrease period. The transition time markers from saturated humidity to continuous decline corresponding to each co-located acquisition point, and the hourly decline rate of relative humidity within the overhead conductor channel during the continuous decline period, are combined to form the conductor periphery drying tendency marker record for that co-located acquisition point. The conductor periphery drying tendency marker records of all co-located acquisition points are summarized according to spatial location to generate a conductor periphery drying tendency marker sequence.

5. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The process of inputting the vegetation combustible state evolution marker sequence and the drying tendency marker sequence around the conductor into the fire hazard spatiotemporal alignment overlay process generates a set of fire hazard alignment overlay relationship records, including: Extract the transition frame time markers from the stable fluctuation state to the continuous decay state of the normalized vegetation index at each co-located collection point from the vegetation combustible state evolution marker sequence, and use the transition frame time markers as the time nodes when the vegetation at the co-located collection point enters the combustible state. Extract the transition time marker from saturated wet state to continuously decreasing state for each co-located acquisition point from the drying tendency marker sequence around the conductor, and use the transition acquisition time marker as the time node when the conductor periphery of the co-located acquisition point enters the dry state. The time nodes when vegetation at the same co-located collection point enters the combustible state and the time nodes when the area around the conductor enters the dry state are compared in chronological order. When the time interval between the time nodes when vegetation enters the combustible state and the time nodes when the area around the conductor enters the dry state is less than the preset fire hazard coupling judgment time window, the co-located collection point is marked as a fire hazard time coupling point. The time nodes when vegetation at the same co-located collection point enters the combustible state are compared with the time nodes when the surrounding area of ​​the conductor enters the dry state. When the time interval between the time nodes when vegetation enters the combustible state and the time nodes when the surrounding area of ​​the conductor enters the dry state is greater than the preset fire hazard coupling judgment time window and less than the preset fire hazard proximity judgment time window, the co-located collection point is marked as a fire hazard time proximity point. The time nodes when vegetation at the same co-located collection point enters the combustible state are compared with the time nodes when the surrounding area of ​​the conductor enters the dry state. When the time interval between the time nodes when vegetation enters the combustible state and the time nodes when the surrounding area of ​​the conductor enters the dry state is greater than the preset fire hazard proximity judgment time window, the co-located collection point is marked as a fire hazard time far away point. The fire hazard point marker type of each co-located collection point is associated with the spatial coordinates of the co-located collection point and stored to form a fire hazard alignment and superposition relationship record. All fire hazard alignment and superposition relationship records of co-located collection points are summarized into a fire hazard alignment and superposition relationship record set. The fire hazard point marker type includes fire hazard time-coupled points, fire hazard time adjacent points, and fire hazard time far away points.

6. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The fire hazard alignment and superposition relationship record set is processed to extract the fire hazard coupling trigger boundary, thereby obtaining fire hazard warning trigger criterion data for power distribution corridors. Based on the fire hazard warning trigger criterion data for power distribution corridors, a fire risk management strategy is generated, including guidance on the location of firebreaks and patrol time window planning, comprising: Extract the frame-by-frame attenuation magnitude of the normalized vegetation index of all co-located collection points marked as fire time-coupled points from the fire hazard alignment and superposition relationship record set, and the hour-by-hour decrease rate of the relative humidity in the overhead conductor channel during the continuous attenuation period. The maximum value of the frame-by-frame attenuation amplitude of the fire risk time-coupled point is extracted as the upper bound reference value of the vegetation yellowing and spreading rate, and the minimum value of the frame-by-frame attenuation amplitude of the fire risk time-coupled point is extracted as the lower bound reference value of the vegetation yellowing and spreading rate. The maximum value of the hourly decrease rate of the fire hazard time-coupled point is extracted as the upper limit reference value of the environmental drying rate, and the minimum value of the hourly decrease rate of the fire hazard time-coupled point is extracted as the lower limit reference value of the environmental drying rate. The lower bound reference value of vegetation yellowing spread rate and the lower bound reference value of environmental drying rate are combined into the lower boundary criterion for triggering fire risk warning in power distribution corridor. The lower boundary criterion for triggering fire risk warning in power distribution corridor is used to trigger a fire risk warning when the frame-by-frame attenuation amplitude of any co-located acquisition point to be monitored reaches the lower bound reference value of vegetation yellowing spread rate and its time-by-time decrease rate simultaneously reaches the lower bound reference value of environmental drying rate. The upper limit reference value of the vegetation yellowing and spreading rate and the upper limit reference value of the environmental drying rate are combined into the upper boundary criterion for triggering the fire risk warning in the power distribution corridor. The upper boundary criterion for triggering the fire risk warning in the power distribution corridor is used to trigger an emergency fire alarm when the frame-by-frame attenuation amplitude of any co-located acquisition point to be monitored reaches the upper limit reference value of the vegetation yellowing and spreading rate and its hour-by-hour decrease rate simultaneously reaches the upper limit reference value of the environmental drying rate. Extract the spatial coordinates of co-located collection points marked as fire time coupling points in the fire hazard alignment and superposition relationship record set, calculate the continuous dense sections of fire time coupling points in the spatial distribution along the overhead power distribution channel, and mark the area outside the boundary of the power distribution corridor covered by the starting coordinates of the continuous dense sections extending outward in the opposite direction of the power distribution line and the ending coordinates extending outward in the positive direction of the power distribution line as the location guide for the opening of fire isolation zones. Extract the time nodes when vegetation enters the combustible state and the time nodes when the area around the conductor enters the dry state from the fire hazard alignment and superposition relationship record set marked as fire hazard time coupling points. Calculate the time overlap interval between the time nodes when vegetation enters the combustible state and the time nodes when the area around the conductor enters the dry state. Convert the time overlap interval into the start and end times of the patrol period window planning to generate the patrol period window planning.

7. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The method further includes: After acquiring the synchronous data packet of the forest and grassland area to be evaluated in the overhead power line channel, the pole tilt sensing data corresponding to each co-located acquisition point along the overhead power line channel is acquired. The pole tilt sensing data includes the offset direction angle and offset distance of the top of the pole relative to the pole base. Based on the offset direction angle and offset distance in the tower tilt sensing data corresponding to each co-located collection point, calculate the three-dimensional displacement vector of the tower top relative to the tower base, and at least decompose the vertical settlement component along the vertical direction and the horizontal displacement component perpendicular to the direction of the distribution line; subtract the vertical settlement component from the design spacing of the overhead conductor to the ground to obtain the actual spacing value of the overhead conductor to the ground considering the influence of tower tilt settlement. Obtain the vegetation canopy height data in the overhead conductor channel corresponding to each co-located collection point, calculate the difference between the actual distance between the overhead conductor and the ground after tower tilt correction and the vegetation canopy height, and obtain the vertical clearance distance between the overhead conductor and the top surface of the vegetation canopy. When the vertical clearance distance is less than the preset safety distance threshold between the conductor and vegetation, a hazard marker for the conductor tree obstacle corresponding to the co-located collection point is generated, the hazard marker for the conductor tree obstacle is added to the fire hazard alignment and superposition relationship record set, and the fire hazard alignment and superposition relationship record set is updated.

8. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The method further includes: After generating the fire risk warning triggering criterion data for the power distribution corridor, historical fire record data of the forest and grassland area to be assessed is obtained. The historical fire record data includes the geographical coordinates of each fire and the date and time of the fire. Spatial matching processing is performed between the geographical coordinates of each fire occurrence in the historical fire record data and the spatial location coordinates of the co-located collection points to determine the historical fire markers corresponding to the co-located collection points where historical fires have occurred. The frame-by-frame decay amplitude of the historical normalized vegetation index of the co-located collection points with historical fire markers before the date and time of the fire occurrence and the hour-by-hour decrease rate of the relative humidity in the historical overhead conductor channel during the continuous decay period are extracted. The historical frame-by-frame attenuation amplitude of co-located collection points with historical fire markers is compared with the lower bound reference value of vegetation withering and spreading rate. When the historical frame-by-frame attenuation amplitude is lower than the lower bound reference value of vegetation withering and spreading rate, the historical frame-by-frame attenuation amplitude is used as the new lower bound reference value of vegetation withering and spreading rate. The historical hourly rate of decay of co-located data collection points with historical fire markers is compared with the lower bound reference value of environmental drying rate. When the historical hourly rate of decay is lower than the lower bound reference value of environmental drying rate, the historical hourly rate of decay is used as the new lower bound reference value of environmental drying rate. The lower bound criterion for triggering fire risk warning in power distribution corridor is updated using the new lower bound reference value of vegetation yellowing and spread rate and the new lower bound reference value of environmental drying rate.

9. The intelligent assessment method for fire risk of power distribution lines in forest and grassland areas according to claim 1, characterized in that, The method further includes: After generating the vegetation combustible state evolution marker sequence, the daily precipitation observation record sequence of the forest and grassland area to be evaluated within the same collection period is obtained. The daily precipitation values ​​in the daily precipitation observation record sequence are compared with the preset effective precipitation threshold. Dates with daily precipitation values ​​lower than the preset effective precipitation threshold are marked as invalid precipitation days, and dates with daily precipitation values ​​higher than the preset effective precipitation threshold are marked as valid precipitation days. Invalid precipitation days are continuously accumulated and counted. When an invalid precipitation day is encountered, the continuous accumulated count is reset to zero and the accumulation starts again to obtain the cumulative number of consecutive invalid precipitation days. The cumulative number of consecutive invalid precipitation days is associated with the time marker of the transition frame when the normalized vegetation index changes from a stable fluctuation state to a continuous decay state. The cumulative number of consecutive invalid precipitation days corresponding to the time marker of the transition frame is extracted as the critical continuous number of days for vegetation drought resistance. The total decrease of the normalized vegetation index in the time series of vegetation canopy greenness characteristics during the continuous decline period is extracted. The ratio of the total decrease to the cumulative number of days with continuous ineffective precipitation is calculated to obtain the vegetation greenness decrease rate per day with ineffective precipitation. The critical duration of drought resistance of vegetation and the vegetation greenness decay rate per unit of ineffective precipitation are used as leading parameters to input into the update process of the vegetation combustible state evolution marker sequence, and the preset stationarity threshold for the stationarity transition detection processing in the time series of vegetation canopy greenness characterization quantities is adjusted.

10. An intelligent assessment system for fire risk of power distribution lines in forest and grassland areas, characterized in that, include: processor; A machine-readable storage medium for storing machine-executable instructions of the processor; The processor is configured to execute the intelligent fire risk assessment method for power distribution lines in forest and grassland areas as described in any one of claims 1 to 9 by executing the machine-executable instructions.

Images