Forest fire spreading prediction method and device, equipment and storage medium

Abstract

The application discloses a mountain fire spreading prediction method and device, equipment and a storage medium, and the method comprises the following steps: acquiring a mountain fire alarm grid of a monitoring area, and performing mountain fire monitoring on the monitoring area based on a high-orbit satellite to obtain fire point distribution in the monitoring area; using a low-orbit satellite to perform staring imaging on the fire points in the monitoring area to obtain satellite condensation imaging video of the fire points in the monitoring area; performing interframe analysis and geographic coding on the satellite condensation imaging video to obtain mountain fire information in the monitoring area; performing position analysis and mountain fire spreading prediction according to the mountain fire information in the monitoring area and the mountain fire alarm grid, determining a predicted position of the mountain fire, and issuing an alarm. The mountain fire spreading is predicted by combining the mountain fire alarm grid information and the real-time monitored mountain fire information, the specific position to which the mountain fire will spread in the mountain fire alarm grid is predicted, timely alarm prevention is made, and the safe operation level of the power grid and the power supply reliability are effectively improved.

Description

Technical Field

[0001] This invention relates to the field of power distribution network safety monitoring technology, and in particular to a method, device, equipment and storage medium for predicting wildfire spread. Background Technology

[0002] In recent years, power outages caused by wildfires have occurred frequently, seriously affecting the safe operation of the power grid. Wildfire monitoring and risk early warning are currently one of the key technologies for power grid disaster prevention and mitigation.

[0003] When wildfires occur near power transmission lines, the high temperatures, smoke, and fly ash generated can easily cause a sharp decrease in the insulation gaps of the transmission lines, triggering power outages. However, current research on wildfire risk warnings mostly focuses on large-scale forest areas, which are too geographically extensive to be applicable to wildfires along power lines. Therefore, it remains impossible to accurately determine the probability of wildfires occurring along power transmission line corridors. Summary of the Invention

[0004] Therefore, it is necessary to propose a method for predicting the spread of wildfires in order to accurately determine the probability of wildfires occurring along power transmission lines.

[0005] To achieve the above objectives, the first aspect of this application provides a method for predicting the spread of wildfires, the method comprising:

[0006] The wildfire alarm grid of the monitoring area is obtained, and wildfire monitoring of the monitoring area is carried out based on high-orbit satellites to obtain the distribution of fire points in the monitoring area;

[0007] Staring imaging of fire points within the monitoring area is performed using low-orbit satellites to obtain satellite staring video of the fire points within the monitoring area;

[0008] Inter-frame analysis and geocoding are performed on the satellite still image video to obtain wildfire information within the monitoring area;

[0009] Based on the wildfire information within the monitoring area and the wildfire alarm grid, location analysis and wildfire spread prediction are performed to determine the predicted location of the wildfire and issue an alarm.

[0010] Furthermore, the method of monitoring wildfires in the monitored area based on high-orbit satellites to obtain the distribution of fire points within the monitored area specifically includes:

[0011] The monitoring images of the monitoring area are acquired based on the high-orbit satellite;

[0012] The solar elevation angle of the pixels in the monitored image, the proportion of non-vegetation pixels and the proportion of cloud pixels within a preset window area are obtained, and the background pixel brightness temperature of the monitored image is obtained using the dynamic window method.

[0013] The preset fire point identification threshold is adaptively corrected based on the solar elevation angle of the pixel, the proportion of non-vegetation pixels, and the proportion of cloud pixels to obtain the target fire point identification threshold.

[0014] The distribution of fire points within the monitoring area is determined by comparing and analyzing the target fire point identification threshold with the background pixel brightness temperature of the monitoring image.

[0015] Furthermore, the step of comparing and analyzing the target fire point identification threshold with the background pixel brightness temperature of the monitoring image to determine the fire point distribution within the monitoring area specifically includes:

[0016] By comparing the target fire point identification threshold with the background pixel brightness temperature of the monitoring image, suspected fire points within the monitoring area are obtained, and the wildfire confidence level of the suspected fire points is calculated.

[0017] When the confidence level of the wildfire meets the preset confidence level conditions, the suspected fire points that meet the confidence level conditions are designated as high-confidence fire points, and the center latitude and longitude of the high-confidence fire points are obtained.

[0018] The distribution of fire points within the monitoring area is obtained based on the center latitude and longitude of the high-confidence fire points.

[0019] Furthermore, the step of performing inter-frame analysis and geocoding on the satellite imagery video to obtain wildfire information within the monitoring area specifically includes:

[0020] Based on the target detection algorithm, the wildfire in the satellite image video is located, and the appearance and motion features of the wildfire in each video frame are extracted.

[0021] Matching and associating corresponding image points between two adjacent video frames in the satellite imagery video to obtain a set of target corresponding image points that have been successfully associated. The motion parameters of each image point in the set of target corresponding image points are added to the target corresponding image point dataset. The set of target corresponding image points is any set of corresponding image points in all sets of corresponding image points.

[0022] The target homonym dataset is updated based on the appearance and motion features of the wildfire in each video frame to obtain the updated target homonym dataset.

[0023] Based on the target homonymous image point dataset, object-side geocoding of the image sequence is performed to obtain the target satellite image;

[0024] Wildfire information is extracted based on the target satellite imagery to obtain the location, area, direction of spread, and maximum spread rate of wildfires within the monitoring area.

[0025] Furthermore, since the grid of the wildfire alarm network includes at least environmental information, the step of performing location analysis and wildfire spread prediction based on wildfire information within the monitoring area and the wildfire alarm network to determine the predicted location of the wildfire and issue an alarm specifically includes:

[0026] Obtain environmental information of the wildfire location in the wildfire alarm grid, the environmental information including at least topography, meteorological data and vegetation type;

[0027] The wildfire spread rate is calculated based on the wildfire information and the environmental information to obtain the wildfire spread rate in each direction;

[0028] Based on the spread rate of the wildfire in various directions and the distance calculation of the wildfire alarm grid, the predicted location of the wildfire is determined and an alarm is issued.

[0029] Furthermore, the grid of the wildfire alarm network also includes power line information. The process of determining the predicted location of the wildfire based on the spread rate of the wildfire in each direction and the distance calculated from the wildfire alarm network, and issuing an alarm, specifically includes:

[0030] Based on the spread rate and the location of the wildfire in the wildfire alarm grid, a predictive analysis is performed to obtain the predicted wildfire grid cells in the wildfire alarm grid, and the predicted grid cells are added to the candidate set.

[0031] Record the ignition time of the grid cells to be ignited in the candidate set, and delete the ignited grid cells from the candidate set to update the candidate set;

[0032] The wildfire information is updated based on the ignited grid cells. When the ignition time of the ignited grid cells is less than the preset simulation time, the step of calculating the wildfire spread rate based on the wildfire information and the environmental information to obtain the spread rate of the wildfire in each direction is repeated.

[0033] Based on the updated candidate set and the transmission line information, probability calculations are performed to determine the predicted location of the wildfire and issue an alarm.

[0034] Furthermore, the transmission line tower information includes at least the corridor buffer zones for each line of the transmission line tower. The step of performing probability calculations based on the updated candidate set and the transmission line tower information to determine the predicted location of a wildfire and issue an alarm specifically includes:

[0035] The probability of wildfire spreading to the area below the target line is calculated and analyzed based on the degree of overlap between all the unburning grids and the corridor buffer zone of the target line in the updated candidate set. The target line is any one of all lines.

[0036] When the probability of a wildfire spreading to the target route is greater than a preset probability threshold, it is determined that a wildfire will occur on the target route, and a wildfire warning is issued.

[0037] To achieve the above objectives, a second aspect of this application provides a wildfire spread prediction device, the device comprising: an information acquisition unit, an information processing unit, and a wildfire prediction unit;

[0038] The information acquisition unit is used to acquire the wildfire alarm grid of the monitoring area, and to conduct wildfire monitoring of the monitoring area based on high-orbit satellites to obtain the distribution of fire points in the monitoring area;

[0039] The information processing unit is used to perform staring imaging on fire points within the monitoring area using low-orbit satellites to obtain satellite staring video of fire points within the monitoring area.

[0040] Inter-frame analysis and geocoding are performed on the satellite still image video to obtain wildfire information within the monitoring area;

[0041] The wildfire prediction unit is used to perform location analysis and wildfire spread prediction based on wildfire information in the monitoring area and the wildfire alarm grid, determine the predicted location of the wildfire, and issue an alarm.

[0042] To achieve the above objectives, a third aspect of this application provides a computer-readable storage medium storing a computer program, characterized in that, when the computer program is executed by a processor, the processor performs the steps of the method described in the first aspect.

[0043] To achieve the above objectives, a fourth aspect of this application provides a computer device including a memory and a processor, characterized in that the memory stores a computer program, which, when executed by the processor, causes the processor to perform the steps of the method described in the first aspect.

[0044] The embodiments of the present invention have the following beneficial effects:

[0045] This invention discloses a method for predicting the spread of wildfires. The method includes: acquiring a wildfire alarm grid for a monitoring area; monitoring wildfires in the monitoring area using high-orbit satellites to obtain the distribution of fire points within the monitoring area; performing staring imaging on the fire points within the monitoring area using low-orbit satellites to obtain satellite staring video of the fire points within the monitoring area; performing inter-frame analysis and geocoding on the satellite staring video to obtain wildfire information within the monitoring area; and performing location analysis and wildfire spread prediction based on the wildfire information and the wildfire alarm grid within the monitoring area to determine the predicted location of the wildfire and issue an alarm. By combining wildfire alarm grid information and real-time monitored wildfire information to predict the spread of wildfires, the method predicts the specific location where the wildfire will spread within the wildfire alarm grid, enabling timely warnings and prevention, and effectively improving the safety operation level of the power grid and the reliability of power supply. Attached Figure Description

[0046] To more clearly illustrate the technical solutions in the embodiments of the present invention or the prior art, the drawings used in the description of the embodiments or the prior art will be briefly introduced below. Obviously, the drawings described below are only some embodiments of the present invention. For those skilled in the art, other drawings can be obtained based on these drawings without creative effort.

[0047] in:

[0048] Figure 1 This is a flowchart illustrating the wildfire spread prediction method according to an embodiment of the present invention.

[0049] Figure 2 This is a schematic diagram of the geographic coding of the target geographic range for satellite video sequence detection in accordance with the present invention;

[0050] Figure 3 This is a structural block diagram of the wildfire spread prediction device according to the present invention;

[0051] Figure 4 This is a diagram showing the internal structure of a computer device in an embodiment of this application. Detailed Implementation

[0052] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0053] For power grids, the transmission and distribution line corridors they focus on only account for a small portion of the forest area. However, most current research on wildfire risk early warning focuses on vast forest areas, which are too geographically large to be applicable to wildfires along power lines, and thus hinders the accurate assessment of the probability of wildfires occurring along transmission and distribution line corridors.

[0054] Based on this, this application provides a method for predicting the spread of wildfires, achieving accurate prediction of the probability of wildfire spread. For details, please refer to [link / reference needed]. Figure 1 , Figure 1 This is a flowchart illustrating the wildfire spread prediction method according to an embodiment of the present invention, specifically including:

[0055] Step 110: Obtain the wildfire alarm grid of the monitoring area, and conduct wildfire monitoring of the monitoring area based on high-orbit satellites to obtain the distribution of fire points in the monitoring area.

[0056] To avoid cumbersome calculations involving wildfires and power transmission lines, information such as power transmission line data, vegetation type, terrain, and alarm levels within the monitoring area are pre-stored in a grid format to improve alarm efficiency. Once a fire is detected or predicted within the grid, information on affected lines or towers and alarm levels can be automatically pushed to power transmission maintenance personnel.

[0057] To address the shortcomings of insufficient timeliness and low resolution in wildfire monitoring, this invention employs high-orbit geostationary satellites (with a spatial resolution typically around 2km) for wildfire monitoring, enabling 24 / 7 continuous monitoring of target areas. For high-orbit satellites, wildfires within a 2km radius are imaged as a single pixel. Therefore, for wildfire monitoring using high-orbit satellites, each wildfire can be considered a single fire point, or image point. Thus, high-orbit satellites can be used to locate high-confidence fire points within the monitoring area, as well as their distribution.

[0058] Step 120: Use low-orbit satellites to perform staring imaging on the fire points in the monitoring area to obtain satellite staring video of the fire points in the monitoring area.

[0059] For low-orbit satellites, due to the improved spatial resolution, the wildfire captured is no longer a single point but an area. Therefore, it is possible to capture images of fire points monitored by high-orbit satellites to obtain video recordings of the wildfire's changes over time, i.e., satellite frozen video.

[0060] Step 130: Perform inter-frame analysis and geocoding on the satellite imagery video to obtain wildfire information within the monitoring area.

[0061] Specifically, each video frame of the satellite imagery is correlated and analyzed, the pixels in each video frame are correlated, and the motion information of each pixel is obtained. The pixels with motion information are geocoded to obtain images with wildfire information, and the wildfire information in the images can be extracted.

[0062] Step 140: Based on the wildfire information and wildfire alarm grid within the monitoring area, perform location analysis and wildfire spread prediction to determine the predicted location of the wildfire and issue an alarm.

[0063] Specifically, the spread of wildfires is predicted based on information about wildfires within the detection area and the location of the ignition point within the wildfire alarm grid. The location of the wildfire spread area is then analyzed against the wildfire alarm grid, which includes information on power transmission lines, vegetation type, topography, and alarm levels, to determine whether the safety of power transmission lines is threatened, and consequently, whether to issue an alarm.

[0064] This invention combines wildfire alarm grid information with real-time monitored wildfire information to predict the spread of wildfires, predicting the specific locations where wildfires will spread within the wildfire alarm grid, so as to issue timely warnings and preventive measures, effectively improving the safety level of the power grid and the reliability of power supply.

[0065] To obtain a more accurate distribution of fire points, step 110 of this embodiment of the invention involves monitoring wildfires in the monitoring area using high-orbit satellites to obtain the distribution of fire points within the monitoring area. This specifically includes:

[0066] Step 1101: Acquire monitoring images of the monitoring area based on high-orbit satellites.

[0067] Specifically, high-resolution infrared imagers on high-orbit geostationary satellites can be used to capture real-time images of the monitored area, thus obtaining monitoring images of the area.

[0068] Step 1102: Obtain the solar elevation angle of the pixels in the monitoring image, the proportion of non-vegetation pixels and cloud pixels within the preset window area, and use the dynamic window method to obtain the background pixel brightness temperature of the monitoring image.

[0069] Specifically, the solar elevation angle of the pixels in the detected image is obtained based on data such as satellite ephemeris, orbital parameters, shooting tilt angle, and shooting time.

[0070] Because there are significant differences in pixel features between vegetation and non-vegetation, vegetation pixels and non-vegetation pixels can be obtained through feature recognition. Alternatively, vegetation cover information can be superimposed beforehand to obtain vegetation pixels and non-vegetation pixels, thereby obtaining the proportion of non-vegetation pixels. It is even possible to determine whether a tree species is combustible and obtain the combustibility information of vegetation.

[0071] When satellites capture images of the ground, clouds can obstruct the view. To eliminate the impact of cloud obstruction on subsequent monitoring results, the monitoring images are classified into cloud pixels and clear-sky pixels. Cloud pixels are images with clouds present, while clear-sky pixels are images where the ground can be directly captured without cloud obstruction. Clouds are identified using the brightness temperature values ​​(brightness temperature) and reflectivity of different channels in the monitoring images, resulting in cloud pixels. The remaining pixels are then classified as clear-sky pixels. The proportion of cloud pixels is then determined based on the number of clear-sky pixels among the cloud pixels.

[0072] Because there is a significant difference between the background pixel brightness temperature of the pixel where the fire occurred and the background pixel brightness temperature of the pixel where no fire occurred, the larger the fire, the greater the difference, and the easier it is to detect the fire. Therefore, the presence or absence of a fire can be determined by acquiring the background pixel brightness temperature of each pixel in the monitoring image.

[0073] Step 1103: Adaptively correct the preset fire point identification threshold based on the pixel solar elevation angle, the proportion of non-vegetation pixels, and the proportion of cloud pixels to obtain the target fire point identification threshold.

[0074] The fire detection threshold is used to determine whether a pixel in the monitored image is a fire point, representing the confidence level of the fire point. To achieve a more accurate fire detection threshold, the preset fire detection threshold is adaptively adjusted based on the obtained pixel solar altitude angle, the proportion of non-vegetation pixels, and the proportion of cloud pixels, resulting in the optimal fire detection threshold for different regions.

[0075] Step 1104: Compare and analyze the target fire point identification threshold with the background pixel brightness temperature of the monitoring image to determine the distribution of fire points within the monitoring area.

[0076] Specifically, the background pixel brightness temperature of each pixel in the monitoring image is compared with the target fire point identification threshold. Pixels that meet the target fire point identification threshold are identified as fire points, thereby obtaining the fire point distribution within the monitoring area.

[0077] To obtain fire points with higher confidence, Step 1104 of this invention involves comparing and analyzing the target fire point identification threshold with the background pixel brightness temperature of the monitoring image to determine the distribution of fire points within the monitoring area. Specifically, this includes:

[0078] Step 11041: Based on the target fire point identification threshold, compare the background pixel brightness temperature of the monitoring image to obtain the suspected fire points in the monitoring area, and calculate the wildfire confidence level of the suspected fire points.

[0079] Step 11042: When the confidence level of the wildfire meets the preset confidence level conditions, the suspected fire points that meet the confidence level conditions are designated as high-confidence fire points, and the center latitude and longitude of the high-confidence fire points are obtained.

[0080] Step 11043: Obtain the distribution of fire points within the monitoring area based on the center latitude and longitude of the high-confidence fire points.

[0081] This invention performs confidence analysis on detected fire points, with four confidence levels: low confidence (low, 1), normal confidence (normal, 2), high confidence (high, 3), and very high confidence (very high, 4). The confidence levels are defined as follows:

[0082] (1) Very high confidence level (4):

[0083]

[0084] In the formula, T7 represents the brightness temperature value of the 3.9μm channel in band 7 of the AHI payload of a high-orbit satellite. T74 represents the background brightness temperature values ​​for all background light in the 3.9μm channel of the AHI payload in high-orbit orbit satellite within the window, while T74 represents the brightness temperature difference between the AHI payload in high-orbit orbit satellites in bands 7 and 14. The background brightness temperature difference values ​​for all AHI payloads in high orbit satellites within the window are defined as follows: landuse = i (i = 0, 2, 3, ..., 7) represents the land use type; Nc represents the number of cloud pixels in the 8 adjacent pixels of the fire pixel; and Nf represents the number of fire pixels in the 8 adjacent pixels of the fire pixel.

[0085] Land use types can be represented as follows:

[0086] Table 1 Definition of Land Use Types

[0087]

[0088] (2) High confidence level (high, 3):

[0089]

[0090] (3) Normal confidence level (2):

[0091] ①

[0092] ②

[0093] ③

[0094] ④

[0095] (4) Low confidence level (low, 1)

[0096] ①

[0097] ②

[0098] Based on this, fire points with high confidence and very high confidence can be classified as high-confidence fire points. Furthermore, to prevent false fire points from influencing monitoring results, a more accurate distribution of false fire points should be obtained. Specifically, false fire point removal mainly includes two parts: removal of fixed high-temperature heat sources and removal of flare effects. The removal of fixed high-temperature heat sources involves creating a false fire point database based on fixed high-temperature heat sources, which can be continuously updated according to the actual wildfire monitoring process to reduce the false alarm rate of fire point identification.

[0099] Due to the observation angle of satellite sensors, sunlight is sometimes reflected back at the boundaries of land and water and cirrus clouds in remote sensing images, producing flares. This results in abnormal brightness temperature values ​​in the mid-infrared band, which is detrimental to the identification of true fire points. This embodiment of the invention also uses the angle between the vector from the Earth's surface to the satellite and the direction of specular reflection to determine fire points. This angle is defined as follows:

[0100] cosθ r =(cos(θ) V cos(θ) s )-sin(θ v sin(θ) s cos(ψ))

[0101] In the formula, θ r Let θ be the angle between the vector from the Earth's surface to the satellite and the direction of specular reflection. V and θ s These are the satellite zenith angle and the solar zenith angle, respectively, and ψ is the relative azimuth angle.

[0102] A pixel is considered to be affected by a solar flare and is removed when all of the following conditions are met:

[0103] θ r <30

[0104] albedo03>0.3

[0105] albedo04>0.3

[0106] By filtering fire point confidence and false fire points, more accurate fire point pixels and fire point distribution can be obtained.

[0107] After acquiring satellite still images of wildfires, wildfire information is obtained by analyzing each video frame. Specifically, step 130 involves performing inter-frame analysis and geocoding on the satellite still images to obtain wildfire information within the monitoring area. This includes:

[0108] Step 1301: Based on the target detection algorithm, locate the wildfire in the satellite image video and extract the appearance and motion features of the wildfire in each video frame.

[0109] Specifically, existing high-performance target detection algorithms are used to extract the motion and appearance features of wildfires or smoke targets in each video frame of satellite imagery, locating wildfire or smoke targets in the video frames, as well as underlying surface information such as vegetation, bare land, farmland, grassland, and buildings around the wildfire. An improved deep learning algorithm can be used to incorporate an attention mechanism into the feature extraction process, enhancing the detector's ability to learn feature representations of weak targets.

[0110] Step 1302: Perform corresponding image point matching and data association between adjacent video frames in the satellite imaging video to obtain a set of successfully associated target corresponding image points. Add the motion parameters of each image point in the target corresponding image point set to the target corresponding image point dataset. The target corresponding image point set is any set of corresponding image points in all corresponding image point sets. Update the target corresponding image point dataset according to the appearance and motion characteristics of the wildfire in each video frame to obtain the updated target corresponding image point dataset.

[0111] Specifically, in the satellite imagery video sequence, the preceding frame of any two adjacent video frames is used as the main frame, and the following frame as the auxiliary frame. For inter-frame matching of corresponding image points, a correlation coefficient matching method is used for initial matching between the main and auxiliary frames. Then, a least-squares image matching method is used to improve matching accuracy. Finally, a random sampling consensus algorithm is used to eliminate mismatched points, resulting in a precise matching result. The motion information of corresponding image points is then assigned to the same motion trajectory based on the obtained precise matching results. Simultaneously, high-confidence and low-confidence detection results are considered to ensure that even detected weak targets can be associated with the trajectory. Trajectory interpolation methods can be used to complete the full trajectory of wildfires or smoke, ensuring the integrity of the predicted trajectory.

[0112] A video frame orientation model is constructed. In this embodiment of the invention, the video frame orientation model is constructed using a terrain-independent rational function model (RFM). By utilizing the parameters of the RFM model solved using a terrain-independent scheme, high fitting accuracy can be achieved without obtaining uniformly distributed real control points, greatly expanding the application scenarios of this invention.

[0113] A frame-by-frame motion estimation strategy is employed to obtain the motion parameters of target image points. In the satellite imagery video sequence, the preceding frame of any two adjacent video frames is used as the master frame, and the following frame as the auxiliary frame. This fully utilizes the orientation parameter information of the satellite imagery video sequence, and based on the video frame orientation model, an inter-frame motion model is constructed to estimate the inter-frame motion parameters frame by frame. Simultaneously, inter-frame motion parameters are introduced based on the auxiliary frame orientation model. After solving for the motion parameters of the current auxiliary frame relative to the master frame, the current auxiliary frame will serve as the master frame for the motion estimation of the next two adjacent frames in the video sequence. To construct the inter-frame motion model for the next two adjacent frames, motion compensation is performed on the orientation parameters of the auxiliary frames, fusing the motion parameters of the current auxiliary frame into the orientation parameters. This eliminates relative attitude errors caused by satellite platform jitter, satellite orbit determination, and attitude measurement errors. Considering the small intersection angle between the master and auxiliary frames, this invention introduces a digital elevation model as an elevation constraint condition, and solves for the motion parameters of the auxiliary frame relative to the master frame through regional network adjustment.

[0114] Step 1303: Perform object geocoding of the image sequence based on the target homonymous image point dataset to obtain the target satellite image.

[0115] Specifically, after inter-frame motion estimation and compensation, the precise geometric relationships between frames in the satellite imagery video are restored. Based on this, object-space geocoding is performed on each frame of the video sequence. See details in [link to relevant documentation]. Figure 2 , Figure 2 This is a schematic diagram of object geocoding for target geographic range detection using satellite video sequences, as implemented in this invention. The geocoding processing steps are as follows:

[0116] S1: Based on the geographic range of the target of interest, generate a regular grid in the object space. The grid resolution can be determined according to the spatial resolution of the video sequence or the requirements. The elevation values ​​of the grid points can be obtained by DEM interpolation (such as bilinear interpolation) or by using the average elevation value.

[0117] S2: Project each ground regular grid point onto all video frames of the restored satellite image video to obtain the image-side projection point.

[0118] S3: Based on the coordinates of the image projection points on the video sequence, perform grayscale resampling and grayscale assignment to obtain a measurable target satellite image for each frame.

[0119] Step 1304: Extract wildfire information based on target satellite imagery to obtain the location, area, direction of spread, and maximum spread rate of wildfires within the monitoring area.

[0120] Measurable satellite imagery of targets can be used to calculate information such as the precise location, area, direction of spread, maximum spread rate, and location and area of ​​vegetation in wildfires.

[0121] This invention proposes a method for predicting the spread area of ​​a wildfire by combining environmental information of the wildfire occurrence. The wildfire alarm grid includes at least environmental information in its grid cells. Therefore, step 140 involves performing location analysis and wildfire spread prediction based on the wildfire information and the wildfire alarm grid within the monitoring area to determine the predicted location of the wildfire and issue an alarm. Specifically, this includes:

[0122] Step 1401: Obtain environmental information about the location of wildfires in the wildfire alarm grid. The environmental information should include at least topography, meteorological data and vegetation type.

[0123] Step 1402: Calculate the wildfire spread rate based on wildfire information and environmental information to obtain the spread rate of the wildfire in each direction.

[0124] This invention combines the surrounding topography of wildfires, meteorological data from meteorological departments, and pre-collected information on underlying surfaces such as vegetation and combustibles, and uses a representative elliptical model to simulate and predict the spread trend of wildfires and the rate of spread of wildfires in various directions.

[0125] For example, in a scenario with no wind and no slope, a wildfire spreads at a constant speed in all directions, resulting in a nearly circular fire area. When weather and terrain conditions remain stable, the spread shape is an elongated ellipse. Given the maximum spread rate and the direction of the maximum spread rate, calculate the spread rate of the wildfire in different directions. The spread rates in other directions can be constructed using the following formula.

[0126]

[0127] In the formula, the initial rate at which the fire spreads in all directions is the same, denoted as V0; and The effect of wind and slope on the rate of wildfire spread; V represents the maximum wildfire spread rate; θ is the angle between the direction of the maximum spread rate and the target (to be calculated) direction; V max This represents the spread rate in other directions.

[0128] Step 1403: Calculate the distance based on the spread rate of the wildfire in each direction and the wildfire alarm grid, determine the predicted location of the wildfire, and issue an alarm.

[0129] The points where fires have already started are considered as current fire sources. Based on the spread rate of the fire source in various directions, the grids that may lead to wildfires are considered as potential burning areas. The shortest path method is used to spread the fire outward grid by grid, thereby achieving the effect of predicting the spread trend of surface fires.

[0130] The wildfire alarm grid in this embodiment of the invention also includes power line information. Therefore, Step 1403 involves calculating the distance based on the wildfire's spread rate in various directions and the wildfire alarm grid to determine the predicted location of the wildfire and issue an alarm. Specifically, this includes:

[0131] Step 14031: Based on the spread rate and the location of the wildfire in the wildfire warning grid, perform predictive analysis to obtain the predicted wildfire grid cells to be burned in the wildfire warning grid, and add the grid cells to the candidate set.

[0132] The ignited points are considered the current fire sources. Based on the wildfire spread rate between different grid cells, the shortest path method is used to calculate the fastest spread time from the current fire source to the surrounding grid cells. Grid cells predicted to be burned are added to the candidate set as potential burning grid cells.

[0133] Step 14032: Record the time when the grid cells to be ignited in the candidate set are ignited, and delete the ignited grid cells from the candidate set to update the candidate set.

[0134] The spread of wildfires is updated in real time. Ignition grids that have been set on fire in the candidate set are removed from the candidate set, and the time when they were set on fire is recorded.

[0135] In addition, once the fire grid has been ignited, the distance between the wildfire and the power transmission lines within the grid is calculated to determine the risk level of the power transmission lines. Early warning levels are assigned in kilometers: less than 1 km is classified as Level 1 risk; 1-2 km as Level 2 risk; and 2-3 km as Level 3 risk. Warnings are issued based on the risk level to achieve early prevention.

[0136] Step 14033: Update the wildfire information based on the ignited grid cells. When the ignition time of the ignited grid cells is less than the preset simulation time, repeat the step of calculating the wildfire spread rate based on the wildfire information and environmental information to obtain the spread rate of the wildfire in each direction.

[0137] The fire source information for ignited grid cells is updated to reflect the new fire information, allowing for real-time prediction of ignited grid cells and updating the predicted location of wildfires. Before updating the fire information, it is necessary to determine whether the ignition time of the ignited grid cell exceeds the preset simulation time. If the ignition time exceeds the preset simulation time, no update is performed, because assuming that the combustible material in a certain area burns for a maximum of 2 hours, continuing to simulate the spread of fire beyond 2 hours is meaningless.

[0138] Step 14034: Perform probability calculations based on the updated candidate set and transmission line information to determine the predicted location of the wildfire and issue an alarm.

[0139] Specifically, the probability of a wildfire spreading to the area below the target route is calculated and analyzed based on the degree of overlap between all the unburning grids in the updated candidate set and the corridor buffer of the target route. The target route is any one of all routes. When the probability of a wildfire spreading to the target route is greater than a preset probability threshold, it is determined that a wildfire will occur on the target route, and a wildfire warning is issued.

[0140] The embodiments of the present invention take into account the influence of the location of transmission lines and environmental factors. By making real-time predictions and updates on the spread of wildfires, more accurate results of wildfire spread are obtained, and alarms are issued at the spread location, thereby achieving the safe operation of the power distribution network.

[0141] This invention also proposes a wildfire spread prediction device; please refer to [link / reference]. Figure 3 , Figure 3 The diagram shows the structure of the wildfire spread prediction device according to the present invention. The device includes: an information acquisition unit 301, an information processing unit 302, and a wildfire prediction unit 303.

[0142] The information acquisition unit 301 is used to acquire the wildfire alarm grid of the monitoring area, and to monitor wildfires in the monitoring area based on high-orbit satellites to obtain the distribution of fire points in the monitoring area.

[0143] The information processing unit 302 is used to perform staring imaging of fire points in the monitoring area using low-orbit satellites to obtain satellite staring video of fire points in the monitoring area.

[0144] Inter-frame analysis and geocoding of satellite still images were performed to obtain wildfire information within the monitoring area.

[0145] The wildfire prediction unit 303 is used to perform location analysis and wildfire spread prediction based on wildfire information and wildfire alarm grids in the monitoring area, determine the predicted location of the wildfire, and issue an alarm.

[0146] The wildfire spread prediction device of this invention predicts the spread of wildfires by combining wildfire alarm grid information and real-time monitored wildfire information. It predicts the specific location in the wildfire alarm grid where the wildfire will spread, so as to make timely warnings and preventive measures, effectively improving the level of power grid safety and power supply reliability.

[0147] Figure 4 An internal structural diagram of a computer device according to one embodiment of the present invention is shown. This computer device can specifically be a terminal or a system. Figure 4As shown, the computer device includes a processor, memory, and network interface connected via a system bus. The memory includes a non-volatile storage medium and internal memory. The non-volatile storage medium stores an operating system and may also store a computer program. When executed by the processor, this computer program causes the processor to perform the steps in the above-described method embodiments. The internal memory may also store a computer program, which, when executed by the processor, causes the processor to perform the steps in the above-described method embodiments. Those skilled in the art will understand that... Figure 4 The structure shown is merely a block diagram of a portion of the structure related to the present application and does not constitute a limitation on the computer device to which the present application is applied. Specific computer devices may include more or fewer components than those shown in the figure, or combine certain components, or have different component arrangements.

[0148] In one embodiment, a computer device is provided, including a memory and a processor, the memory storing a computer program that, when executed by the processor, causes the processor to perform the steps in the above method embodiments.

[0149] In one embodiment, a computer-readable storage medium is provided storing a computer program that, when executed by a processor, causes the processor to perform the steps in the above method embodiments.

[0150] Those skilled in the art will understand that all or part of the processes in the methods of the above embodiments can be implemented by a computer program instructing related hardware. The program can be stored in a non-volatile computer-readable storage medium, and when executed, it can include the processes of the embodiments of the above methods. Any references to memory, storage, databases, or other media used in the embodiments provided in this application can include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM), or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. By way of illustration and not limitation, RAM is available in various forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), dual data rate SDRAM (DDRSDRAM), enhanced SDRAM (ESDRAM), synchronous link DRAM (SLDRAM), RAMbus direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and RAMbus dynamic RAM (RDRAM), etc.

[0151] The technical features of the above embodiments can be combined in any way. For the sake of brevity, not all possible combinations of the technical features in the above embodiments are described. However, as long as there is no contradiction in the combination of these technical features, they should be considered to be within the scope of this specification.

[0152] The above embodiments merely illustrate several implementation methods of this application, and while the descriptions are relatively specific and detailed, they should not be construed as limiting the scope of this patent application. It should be noted that those skilled in the art can make various modifications and improvements without departing from the concept of this application, and these all fall within the protection scope of this application. Therefore, the protection scope of this patent application should be determined by the appended claims.

Claims

1. A method for predicting the spread of wildfires, characterized in that, The method includes: The wildfire alarm grid of the monitoring area is obtained, and wildfire monitoring of the monitoring area is carried out based on high-orbit satellites to obtain the distribution of fire points in the monitoring area. The wildfire alarm grid includes power line information, vegetation type, topography and alarm level. Staring imaging of fire points within the monitoring area is performed using low-orbit satellites to obtain satellite staring video of the fire points within the monitoring area; Inter-frame analysis and geocoding are performed on the satellite image video to obtain wildfire information within the monitoring area, wherein the wildfire information includes wildfire location, wildfire area, wildfire spread direction and maximum wildfire spread rate; Based on the wildfire information within the monitoring area and the wildfire alarm grid, location analysis and wildfire spread prediction are performed to determine the predicted location of the wildfire and issue an alarm; Specifically, the step of performing inter-frame analysis and geocoding on the satellite imagery to obtain wildfire information within the monitoring area includes: Based on the target detection algorithm, the wildfire in the satellite image video is located, and the appearance and motion features of the wildfire in each video frame are extracted. Matching and associating corresponding image points between two adjacent video frames in the satellite imagery video to obtain a set of target corresponding image points that have been successfully associated. The motion parameters of each image point in the set of target corresponding image points are added to the target corresponding image point dataset. The set of target corresponding image points is any set of corresponding image points in all sets of corresponding image points. The target homonym dataset is updated based on the appearance and motion features of the wildfire in each video frame to obtain the updated target homonym dataset. Based on the target homonymous image point dataset, object-side geocoding of the image sequence is performed to obtain the target satellite image; Wildfire information is extracted based on the target satellite imagery to obtain the location, area, direction of spread, and maximum spread rate of wildfires within the monitoring area.

2. The method according to claim 1, characterized in that, The method of monitoring wildfires in the monitored area based on high-orbit satellites to obtain the distribution of fire points within the monitored area specifically includes: The monitoring images of the monitoring area are acquired based on the high-orbit satellite; The solar elevation angle of the pixels in the monitored image, the proportion of non-vegetation pixels and the proportion of cloud pixels within a preset window area are obtained, and the background pixel brightness temperature of the monitored image is obtained using the dynamic window method. The preset fire point identification threshold is adaptively corrected based on the solar elevation angle of the pixel, the proportion of non-vegetation pixels, and the proportion of cloud pixels to obtain the target fire point identification threshold. The distribution of fire points within the monitoring area is determined by comparing and analyzing the target fire point identification threshold with the background pixel brightness temperature of the monitoring image.

3. The method according to claim 2, characterized in that, The step of comparing and analyzing the target fire point identification threshold with the background pixel brightness temperature of the monitoring image to determine the fire point distribution within the monitoring area specifically includes: By comparing the target fire point identification threshold with the background pixel brightness temperature of the monitoring image, suspected fire points within the monitoring area are obtained, and the wildfire confidence level of the suspected fire points is calculated. When the confidence level of the wildfire meets the preset confidence level conditions, the suspected fire points that meet the confidence level conditions are designated as high-confidence fire points, and the center latitude and longitude of the high-confidence fire points are obtained. The distribution of fire points within the monitoring area is obtained based on the center latitude and longitude of the high-confidence fire points.

4. The method according to claim 1, characterized in that, The wildfire alarm grid includes at least environmental information. The process of performing location analysis and wildfire spread prediction based on wildfire information within the monitoring area and the wildfire alarm grid to determine the predicted location of a wildfire and issue an alarm specifically includes: Obtain environmental information of the wildfire location in the wildfire alarm grid, the environmental information including at least topography, meteorological data and vegetation type; The wildfire spread rate is calculated based on the wildfire information and the environmental information to obtain the wildfire spread rate in each direction; Based on the spread rate of the wildfire in various directions and the distance calculation of the wildfire alarm grid, the predicted location of the wildfire is determined and an alarm is issued.

5. The method according to claim 4, characterized in that, The wildfire alarm grid also includes power line information. The process of determining the predicted location of the wildfire based on the spread rate of the wildfire in each direction and the distance calculated from the wildfire alarm grid, and issuing an alarm, specifically includes: Based on the spread rate and the location of the wildfire in the wildfire alarm grid, a predictive analysis is performed to obtain the predicted wildfire grid cells in the wildfire alarm grid, and the predicted grid cells are added to the candidate set. Record the ignition time of the grid cells to be ignited in the candidate set, and delete the ignited grid cells from the candidate set to update the candidate set; The wildfire information is updated based on the ignited grid cells. When the ignition time of the ignited grid cells is less than the preset simulation time, the step of calculating the wildfire spread rate based on the wildfire information and the environmental information to obtain the spread rate of the wildfire in each direction is repeated. Based on the updated candidate set and the transmission line information, probability calculations are performed to determine the predicted location of the wildfire and issue an alarm.

6. The method according to claim 5, characterized in that, The transmission line tower information includes at least the corridor buffer zones for each line of the transmission line tower. The process of performing probability calculations based on the updated candidate set and the transmission line tower information to determine the predicted location of a wildfire and issuing an alarm specifically includes: The probability of wildfire spreading to the area below the target line is calculated and analyzed based on the degree of overlap between all the unburning grids and the corridor buffer zone of the target line in the updated candidate set. The target line is any one of all lines. When the probability of a wildfire spreading to the target route is greater than a preset probability threshold, it is determined that a wildfire will occur on the target route, and a wildfire warning is issued.

7. A wildfire spread prediction device, characterized in that, The device includes: an information acquisition unit, an information processing unit, and a wildfire prediction unit; The information acquisition unit is used to acquire the wildfire alarm grid of the monitoring area, and to monitor wildfires in the monitoring area based on high-orbit satellites to obtain the distribution of fire points in the monitoring area. The wildfire alarm grid includes power transmission line information, vegetation type, topography and alarm level. The information processing unit is used to perform staring imaging on fire points within the monitoring area using low-orbit satellites to obtain satellite staring video of the fire points within the monitoring area; and to perform inter-frame analysis and geocoding on the satellite staring video to obtain wildfire information within the monitoring area, wherein the wildfire information includes wildfire location, wildfire area, wildfire spread direction, and wildfire maximum spread rate. The wildfire prediction unit is used to perform location analysis and wildfire spread prediction based on wildfire information in the monitoring area and the wildfire alarm grid, determine the predicted location of the wildfire, and issue an alarm. The information processing unit is further configured to: locate wildfires in the satellite imagery video based on a target detection algorithm; extract the appearance and motion features of wildfires in each video frame of the satellite imagery video; perform homonymous pixel matching and data association between adjacent video frames in the satellite imagery video to obtain a set of successfully associated target homonymous pixels; add the motion parameters of each pixel in the target homonymous pixel set to the target homonymous pixel dataset, wherein the target homonymous pixel set is any set of homonymous pixels in all homonymous pixel sets; update the target homonymous pixel dataset according to the appearance and motion features of the wildfire in each video frame to obtain an updated target homonymous pixel dataset; perform object-space geocoding of the image sequence according to the target homonymous pixel dataset to obtain target satellite images; and extract wildfire information based on the target satellite images to obtain the location, area, direction of wildfire spread, and maximum spread rate of wildfire within the monitoring area.

8. A computer-readable storage medium storing a computer program, characterized in that, When the computer program is executed by a processor, the processor performs the steps of the method as described in any one of claims 1 to 6.

9. A computer device, comprising a memory and a processor, characterized in that, The memory stores a computer program that, when executed by the processor, causes the processor to perform the steps of the method as described in any one of claims 1 to 6.

Images