CN-121999568-A - Multi-source perception early fire point judging method and system

CN121999568ACN 121999568 ACN121999568 ACN 121999568ACN-121999568-A

Abstract

The invention discloses a multi-source perception early fire point judging method and system, and relates to the technical field of fire monitoring and early warning. The method comprises the steps of dividing a three-dimensional grid, mapping visible light, infrared, smoke, gas and temperature and humidity according to imaging geometry and sensor positions, adding a time index to form multi-source data, calculating a hillside wind porch-channel coupling coefficient according to elevation difference, range and dominant wind direction and wind speed amplification amount, extracting brightness fluctuation amount and temperature polymerization amount to obtain apparent heat abnormal amount, combining smoke gas increment and drying and heating indexes to obtain environment abnormal amount, and determining early fire point area positions according to comprehensive fire point judgment amount and threshold connectivity. By uniformly fusing multisource perception data on a three-dimensional grid, introducing wind porch paths of coupling coefficients of mountain regions to describe the relationship between the terrain and the wind field, and constructing comprehensive fire point judgment quantity of the synergistic effect of vision, heat radiation and environment abnormality, fine positioning and stable identification of early fire points in a complex mountain region environment are realized.

Inventors

ZHAO ZHENHUA
YANG YANG
WANG JIANGTAO
LI XUEBIN
HU ZIQIANG

Assignees

国网河北省电力有限公司保定供电分公司

Dates

Publication Date: 20260508
Application Date: 20251211

Claims (10)

1. A method for determining early fire points by multi-source perception, comprising the steps of: Dividing a three-dimensional grid according to the digital elevation model, and mapping image data, smoke concentration, gas concentration and temperature and humidity data to the grid according to an imaging geometric relationship to form multi-source data; generating terrain channel direction information according to elevation difference and local range, and generating a mountain wind porch-channel coupling coefficient according to dominant wind direction and wind speed amplification; constructing brightness fluctuation quantity and temperature polymerization quantity according to the continuous moment visible light brightness change and the infrared radiation change to form vision and heat radiation abnormal quantity; Constructing a drying and heating index according to the relation between smoke, gas concentration increment and temperature and humidity change, and combining the drying and heating index into an environment comprehensive abnormal quantity; And constructing a comprehensive fire point judgment amount according to the two abnormal amounts and the wind porch paths of coupling coefficients of the mountain land, classifying according to a threshold value, performing connectivity analysis, and determining an early fire point area and a spatial position.
2. The method of claim 1, wherein dividing the three-dimensional grid according to the digital elevation model to form the multi-source data comprises: Selecting discrete intervals of the grids in the horizontal direction and the vertical direction according to the elevation change range of the monitoring area and the ground coverage condition; Determining weights according to the space distances between the center of the grid and the projection points of the imaging equipment and the space distances between the smoke and the mounting points of the gas sensor; and carrying out space distribution on the visible light brightness, the infrared radiation intensity, the smoke concentration and the gas concentration according to the weight, so that the numerical value of the multi-source observation on each grid at the same time is determined by the distance relation and the topographic relation.
3. The method of claim 2, wherein the forming of the mountain wind porch-way coupling coefficients comprises: constructing slope information according to elevation differences of the surface grids at adjacent positions; constructing topographic extremely poor information according to the maximum elevation and the minimum elevation difference in the local window; determining wind direction components and wind speed strengthening amounts according to dominant wind directions and wind speed amplifying amounts obtained by meteorological observation; And forming a hillside wind porch-channel coupling coefficient according to the combination relation of the gradient information and the dominant wind direction and the topographic extremely poor information and the wind speed strengthening amount, so that the grids in the valley channel, saddle or high wind speed area are different from the grids in the leeward gradient area in value on the coefficient.
4. The method of claim 1, wherein the forming of the visual and thermal radiation anomaly values comprises: constructing a brightness difference amount according to the visible brightness change under continuous time; constructing brightness fluctuation quantity according to the accumulated change of multi-frame brightness difference quantity in a set time window; constructing a temperature variation according to the infrared radiation variation in continuous time; constructing a temperature polymerization amount according to a polymerization result of the temperature variation in the window; visual and thermal radiation abnormal quantities are formed according to a nonlinear combination relation between brightness fluctuation quantity and temperature aggregation quantity, and values are assigned to the surface grid in a pixel range which is identified in advance through flame candidate areas.
5. The method for determining early fire of multi-source perception according to claim 4, wherein the forming of the environment complex abnormal quantity comprises: generating smoke increment according to the smoke concentration difference value between the current time and the reference time; generating a gas increment according to the gas concentration difference value between the current time and the reference time; generating a drying and heating index according to the temperature rising amplitude and the relative humidity reducing degree; and constructing environment comprehensive abnormal quantity according to a polynomial combination relation between a time accumulation result of smoke increment and gas increment in a time window and a drying and heating index, so that the grids of the smoke and gas continuously rising and in a high-temperature low-humidity environment have higher values on the environment comprehensive abnormal quantity.
6. The method for determining the early fire point by multi-source perception according to claim 5, wherein the construction of the comprehensive fire point determination amount comprises: Nonlinear amplification is carried out on the vision and heat radiation abnormal quantity and the environment comprehensive abnormal quantity according to the fixed exponential relation of the vision and heat radiation abnormal quantity and the environment comprehensive abnormal quantity; the nonlinear amplification is enhanced according to the wind porch-channel coupling coefficient of mountain land; and normalizing according to the sum of the two abnormal quantities and the coupling coefficient of the hilly region wind porch channels to form a comprehensive fire point judgment quantity which simultaneously comprises visual information, environment information and terrain wind field information.
7. The method of claim 6, wherein the step of determining the grid comprises: determining a high confidence fire point grid set according to the relation between the comprehensive fire point judgment quantity and the high threshold value; Determining a suspicious fire point grid set according to the relation between the comprehensive fire point judgment quantity and the high threshold value and the low threshold value; determining a non-fire point grid set according to the relation between the comprehensive fire point judgment quantity and the low threshold value; And carrying out connectivity analysis according to the adjacent relation on grid indexes among the high-confidence fire point grids at the same time, and determining the outline of the early fire point area with continuous space.
8. The method of claim 1, wherein determining the spatial location of the early fire region comprises: Calculating the boundary positions of the area in the horizontal direction and the vertical direction according to the grid index set in the early fire point area obtained by the connectivity analysis; obtaining a plane coordinate range and a height range of the early fire point area according to the grid size and the reference coordinate position; and determining the central position coordinates of the early fire point areas according to the middle position of the grid index and correlating with the corresponding time index.
9. The method for judging early-stage fire points by multi-source perception according to claim 1, wherein the calculation result records under each time according to the multi-source data and the comprehensive fire point judgment amount comprise fire point judgment records of time indexes, grid indexes, two abnormal quantities, a hilly wind porch-channel coupling coefficient and grading judgment results; And archiving and storing the fire judgment records according to the monitoring period, and establishing a retrievable index according to the time sequence and the space range so as to recheck and track the early fire judgment process according to the historical records during subsequent analysis.
10. A multi-source aware early fire decision system for implementing a multi-source aware early fire decision method according to any one of claims 1-9, comprising: the three-dimensional grid mapping module is used for dividing a three-dimensional grid according to the digital elevation model, mapping image data, smoke concentration, gas concentration and temperature and humidity data to the grid according to the imaging geometric relationship and the sensor position, and forming multi-source data; The wind gallery coupling module is used for generating terrain channel direction information according to elevation difference and local range and generating a mountain wind porch-channel coupling coefficient according to dominant wind direction and wind speed amplification; the optothermal anomaly module is used for constructing brightness fluctuation quantity and temperature polymerization quantity according to the visible light brightness change and the infrared radiation change at continuous moments to form vision and heat radiation anomaly quantity; the environment anomaly module is used for constructing a drying and heating index according to the relation between the smoke, the gas concentration increment and the temperature and humidity change, and combining the drying and heating index into an environment comprehensive anomaly; And the comprehensive judgment module is used for constructing comprehensive fire point judgment quantity according to the two abnormal quantities and the wind porch paths of coupling coefficients of the mountain land, classifying according to the threshold value, performing connectivity analysis and determining early fire point areas and space positions.

Description

Multi-source perception early fire point judging method and system Technical Field The invention relates to the technical field of fire monitoring and early warning, in particular to a multi-source perception early fire point judging method and system. Background The existing mountain early fire monitoring is mostly judged by a single sensing means, and typical modes comprise visible light flame identification based on video images, temperature anomaly detection based on infrared imaging, concentration threshold judgment based on smoke or flammable gas sensors and the like. The method generally regards mountain environments as plane areas, does not distinguish the influence of complex terrains on observation results, and only performs feature extraction and threshold judgment on single-class perception data. For example, a flame detection method based on visible light photographing recognizes suspected fires by analyzing color changes and brightness jumps, but is easily interfered by terrain shielding, shadow changes and sunlight reflection in a mountain environment, resulting in a large number of erroneous judgment. The temperature anomaly detection method based on infrared radiation generally utilizes a fixed threshold value to identify a hot spot area, but the slope difference of mountain land and uneven ground surface light reception can cause a large number of non-fire source hot spots to be mistakenly identified as fire spots. The monitoring mode based on the smoke sensor or the gas sensor is limited by the distribution density and wind direction change, and smoke plumes often show phenomena of being guided, detained or flowing around in complex valleys, so that single-point concentration change shows highly unstable characteristics under different terrain conditions. In the prior art, although partial schemes attempt to improve reliability through multi-source fusion, most schemes simply superimpose or use abnormal results of different sensors in parallel, lack uniform three-dimensional space modeling, and cannot correspond visible light, infrared, smoke, gas and meteorological features on a common space-time scale. In addition, the influence of mountain ventilation channels on a fire rising path is not generally considered in the existing method, and a topographic wind guiding effect is not introduced in the judging process, so that early fire point identification hysteresis in high wind sensing areas such as valley areas, saddle areas and the like is caused, and even complete failure is caused. Particularly when noise or local abnormality exists in multi-source data, due to the lack of nonlinear coupling modeling and space-time consistency analysis among multiple elements, the existing scheme is difficult to distinguish local brightness disturbance, high Wen Debiao, humidity deviation or false signals caused by non-fire source smoke, so that the stability and accuracy of early fire point detection are limited. In summary, the prior art mainly has the defects that multisource perception is not uniformly mapped to a three-dimensional grid, coupling parameters based on terrain and wind fields are lacked, vision and environment abnormality are not uniformly modeled, comprehensive judgment quantity is not built due to multisource abnormality, and the like in early stage judgment of mountain fire, and cannot meet the requirements of accurate positioning and stable recognition of early fire points in complex mountain scenes. Disclosure of Invention Based on the above-mentioned drawbacks of the prior art, the present invention is directed to a multi-source sensing early fire judgment method and system, so as to solve the above-mentioned technical problems. In order to achieve the purpose, the invention provides the following technical scheme that the multi-source perception early fire point judging method comprises the following steps: dividing a three-dimensional grid according to the digital elevation model, and mapping image data, smoke concentration, gas concentration and temperature and humidity data to the grid according to an imaging geometric relationship and a sensor position to form multi-source data; generating terrain channel direction information according to elevation difference and local range, and generating a mountain wind porch-channel coupling coefficient according to dominant wind direction and wind speed amplification; constructing brightness fluctuation quantity and temperature polymerization quantity according to the continuous moment visible light brightness change and the infrared radiation change to form vision and heat radiation abnormal quantity; Constructing a drying and heating index according to the relation between smoke, gas concentration increment and temperature and humidity change, and combining the drying and heating index into an environment comprehensive abnormal quantity; And constructing a comprehensive fire point judgment amount according to the two abnormal amount