An active cooling fire scene panoramic monitoring system

Abstract

This invention relates to the field of panoramic fire monitoring technology, specifically an active cooling panoramic fire monitoring system. The system includes a time-series alignment processing module for synchronous data generation and temperature image recording; a flame concentration identification module for extracting concentrated flame areas based on color and spatiotemporal characteristics; a smoke approach identification module for determining smoke obstruction approach areas based on image change trends; a heat zone extraction module for combining temperature data to eliminate obstructions and lock high-heat concentration areas; and a flow direction switching module for adjusting valves based on these parameters and temperature rise points to output monitoring execution results. This invention enhances the consistency of fire information correspondence by establishing multi-dimensional time-series correlations and filtering out data disturbed by air curtains. It identifies flame boundaries and heat accumulation ranges by combining color brightness characteristics and smoke trends, reducing obstruction and airflow interference. It switches cooling flow directions based on the relationship between continuous circumferential temperature rise and high-heat orientation, improving the circumferential heat dissipation balance of the spherical cover to alleviate heat accumulation and ensure continuous and stable high-temperature panoramic monitoring.

Description

Technical Field

[0001] This invention relates to the field of panoramic fire monitoring technology, and in particular to an active cooling panoramic fire monitoring system. Background Technology

[0002] The field of panoramic fire monitoring technology includes fire environment visual monitoring technology, fire scene image acquisition technology, fire dynamic analysis technology, and high-temperature monitoring technology. Over the years, with the widespread application of video surveillance equipment, computer vision-based fire monitoring has gradually become an important means of fire scene analysis. Its core content mainly includes fire scene video acquisition, smoke and fire area identification, fire spread process recording, and continuous monitoring of fire scene status. Existing fire scene monitoring equipment has poor stability in high-temperature environments and short continuous working time. It also has problems such as limited field of view, insufficient imaging quality, and difficulty in covering the entire fire scene area from a fixed perspective. Moreover, it is mostly focused on smoke and fire identification and lacks the ability to continuously analyze changes in flame areas and heat release processes, making it difficult to meet the long-term and panoramic monitoring needs in fire scene environments.

[0003] Among them, the active cooling panoramic fire monitoring system refers to a panoramic fire monitoring system that combines an active cooling structure with a panoramic camera PTZ to continuously monitor the fire environment. It mainly addresses the problems of insufficient heat resistance, limited field of view coverage, and poor stability of continuous video acquisition of monitoring equipment under high-temperature conditions at fire sites. It typically uses a coolant storage container, high-temperature resistant delivery pipelines, and directional spray nozzles to deliver coolant to the outer surface of the panoramic camera PTZ and its surrounding area. The liquid coverage reduces heat accumulation, and the heat sink structure facilitates heat conduction. The panoramic camera PTZ with a rotating structure achieves large-area image acquisition through horizontal rotation and pitch adjustment. Combined with changes in flame area, flame outline, and brightness distribution in the continuous video, the system continuously records and analyzes the fire status.

[0004] The existing panoramic fire monitoring process mainly relies on fixed cooling paths and continuous video content to determine the fire status. Under the condition of continuous changes in heat distribution inside the fire, uneven heat dissipation is likely to occur at different circumferential positions of the dome. When local areas are in a high-temperature state for a long time, the temperature difference on the surface of the dome gradually increases, which can easily affect the stability of continuous monitoring. At the same time, under the environment of smoke diffusion, liquid spraying and airflow in the fire, local areas of the image are prone to brightness fluctuations, edge weakening and grayscale changes. For example, when smoke obscures the flame area, it is difficult to accurately distinguish the actual flame boundary from the environmental interference area based solely on the image content, which can easily cause deviations in the judgment of hot zones and lags in cooling adjustment. Summary of the Invention

[0005] The purpose of this invention is to overcome the shortcomings of existing technologies and propose an active cooling panoramic monitoring system for fire scenes.

[0006] To achieve the above objectives, the present invention adopts the following technical solution: an active cooling fire scene panoramic monitoring system, the system comprising: The timing alignment processing module acquires continuous images, image acquisition time, temperature readings, flow meter readings, cooling medium readings, electric valve readings, and air curtain status. It collects related data groups according to the sampling time corresponding to the preset time period, removes related data groups affected by airflow based on the air curtain ejection interval, and writes them into the same sequence to obtain synchronous recording of temperature images. The flame concentration identification module, based on the continuous images and acquisition time recorded synchronously in the temperature image, calls the pixel red, green and blue and brightness readings, compares the red and green, red and blue bias and brightness shift, marks the candidate flame pixels, merges them according to spatial adjacency and extracts the continuous outer edge extension area frame by frame to obtain the flame concentration area. The smoke approach recognition module, based on the temperature image synchronously recorded, calls grayscale, edge and low contrast ratio to identify the smoke area, combines the air curtain coverage range and the ejection direction to filter out airflow interference, and compares the area with the same moving direction and approaching direction at any time to obtain the smoke obstruction approach area. The heat zone extraction module, based on the flame concentration area and the smoke shielding approach area, calls the surface temperature reading of the transparent cover, collects the flame distribution and identifies the longitudinal and lateral expansion areas, corresponding to the temperature concentration areas, and excludes the parts that fall into the area with the same direction, thus obtaining the high heat concentration area. The flow direction switching module calls the temperature screen synchronously recorded and the high heat concentration area, continuously compares the temperature readings at adjacent sampling times in the same circumferential position, identifies the circumferential position that is continuously heating up and corresponds to high heat, compares it with the preset flow channel starting direction and adjusts the valve to obtain the panoramic monitoring execution result.

[0007] The temperature display synchronously records information including display time correspondence, temperature correlation, air curtain status, flow rate correspondence, valve status, and circumferential temperature distribution. The flame concentration area includes flame coverage, flame edge shape, flame extension shape, flame continuity distribution, and flame brightness distribution. The smoke obstruction approach area includes smoke diffusion range, smoke obstruction boundary, smoke movement direction, smoke concentration distribution, and smoke approach area marker. The high heat concentration area includes heat accumulation range, longitudinal heat expansion area, lateral heat expansion area, temperature concentration distribution, and heat zone boundary information. The panoramic monitoring execution results include flow direction switching status, valve control status, cooling medium flow status, heat concentration location, and cooling adjustment results.

[0008] The working principle and advantages of this invention are as follows: In this invention, a unified temporal correlation is established by continuous images, temperature readings, flow information, and valve status. Data content affected by disturbances is filtered out based on the air curtain ejection status to enhance the consistency of fire scene information. The flame expansion boundary and heat accumulation range are identified by combining red bias characteristics, brightness shift characteristics, and smoke approach direction, reducing the interference of smoke obstruction and airflow fluctuations on hot zone positioning. The cooling medium flow direction is switched according to the correspondence between the circumferential continuous heating state and the high heat concentration location, improving the circumferential heat dissipation uniformity of the spherical cover, alleviating the heat accumulation phenomenon caused by local continuous heating, and ensuring the continuous and stable panoramic monitoring process under high temperature environment. Attached Figure Description

[0009] Figure 1 This is a system flowchart of the present invention; Figure 2 This is a flowchart illustrating the acquisition process of the timing alignment processing module of the present invention. Figure 3 This is a flowchart illustrating the acquisition process of the flame centralized identification module of the present invention. Figure 4 This is a flowchart illustrating the acquisition process of the flue gas approach identification module of the present invention. Figure 5 This is a flowchart illustrating the acquisition process of the hot zone extraction module of the present invention. Figure 6 This is a flowchart of the flow direction switching module of the present invention. Detailed Implementation

[0010] The following detailed explanation illustrates the specific implementation methods: The basic implementation examples are as follows: Figure 1 As shown: An active cooling fire scene panoramic monitoring system, comprising: The timing alignment processing module acquires continuous images captured by the panoramic PTZ camera, the image acquisition time, the PTZ drain temperature reading, the PTZ transparent cover surface temperature reading, the circumferential temperature readings of the annular cooling channel, the cooling medium inlet temperature reading, the cooling medium outlet temperature reading, the flow meter reading, the electric valve opening reading, and the corresponding sampling time records for each reading. It also records the air compressor outlet status and the annular outlet pipe spray status in the annular dust removal air curtain equipment. According to the preset sampling time periods, it matches the image acquisition time with the sampling time of each reading, and aligns the images, PTZ drain temperature reading, PTZ transparent cover surface temperature reading, and annular cooling channel temperature readings within the same sampling time period. Temperature readings at each circumferential position of the flow channel, inlet temperature readings of the cooling medium, outlet temperature readings of the cooling medium, flow meter readings, electric valve opening readings, and sampling time records are collected into associated data groups. The air curtain ejection duration range is determined based on the air compressor outlet state and the annular outlet pipe ejection state at the continuous sampling time. Associated data groups whose image acquisition time falls within the air curtain ejection duration range and whose corresponding airflow ejection state is removed. The retained associated data groups are written into the same time-series data sequence according to the sampling time order, so that the continuous image, image acquisition time, temperature readings, flow meter readings, electric valve opening readings, and air curtain state are arranged synchronously according to the sampling time, resulting in a synchronous temperature image record. The temperature display is recorded synchronously, including display time information, temperature correlation information, air curtain status information, flow rate information, valve status information, and circumferential temperature distribution information.

[0011] Please see Figure 2 Specifically, the timing alignment processing module includes: The information access submodule acquires continuous images captured by the panoramic PTZ camera, the image acquisition time, the PTZ cover drainage temperature reading, the PTZ cover transparent surface temperature reading, the temperature readings at each circumferential position of the annular cooling channel, the cooling medium inlet temperature reading, the cooling medium outlet temperature reading, the flow meter reading, the electric valve opening reading, and the sampling time record corresponding to each reading. It then organizes the correspondence between the images and readings according to the sampling time record, generating a set of sampling information; specifically as follows: The system acquires continuous images and various parameters from the panoramic PTZ camera. The processor calls the video stream frame data from the PTZ camera's transmission interface in real time. Specifically, the image data is acquired at a frequency of 60 frames per second using a built-in CMOS image sensor. The frame index number corresponding to each frame in the continuous image is bound to the generation time generated by the system's internal high-precision hardware real-time clock (RTC). Values ​​fed back from the PTZ cover drainage temperature sensor (specifically a PT100 platinum resistance temperature sensor) are read, such as... Degrees Celsius, read the temperature of the surface sensor (using a patch-type K-type thermocouple) of the transparent dome at the corresponding frame time, as shown below. Temperature data were obtained from temperature sensors located at four circumferential positions in the annular cooling channel, at degrees Celsius. Celsius Celsius Celsius and Degrees Celsius, retrieve the Kelvin temperature conversion value or Celsius reading at the cooling medium inlet, such as... Temperature in degrees Celsius, and the reading at the outlet, such as The instantaneous flow rate in the pipeline is obtained by measuring the flow meter signal line (specifically, by calling the 4-20mA analog signal output by the electromagnetic flow meter). The flow rate is cubic meters per hour. The opening value is determined by reading the percentage signal from the potentiometer fed back by the electric valve. The physical quantity information fed back by all the above sensors is extracted along with their respective sampling stamps in the data acquisition board, for the time of image acquisition. With the time of reading sampling Perform correlation mapping, setting the image acquisition time as the reference point, and searching the database for the sampling times with the smallest absolute difference from it. If the sampling time deviation of the drainage temperature corresponding to a certain image is... If the time interval is milliseconds, the temperature reading will be attributed to the current frame rate. If the minimum deviation is found to be greater than the maximum allowable delay threshold, then... If the time is less than a millisecond, it is determined that the sensor's current period data is lost. Linear interpolation is used to fill in the missing data using historical data from before and after the loss. Specifically, if the first frame of data is missing, backward difference is used for estimation; if the last frame is missing, forward difference is used and the estimated value is recorded. The PT100 sensor's accuracy conforms to the IEC60751 Class A standard. The electromagnetic flowmeter's accuracy conforms to ISO 9104 standard. (Temperature: Celsius) By repeatedly performing the aforementioned timing stamping and matching operation on each frame, a multi-dimensional data tuple consisting of the image, temperature at each location, flow rate, and opening degree is established. The original pulse signals or analog signals from all sensors within the same sampling period are converted into digital quantities by an analog-to-digital converter and arranged linearly according to the timing axis to generate a set of sampling information.

[0012] The time-period aggregation submodule, based on the sampling information set, calls the screen acquisition time, the sampling time record corresponding to each reading, and the preset sampling time period. It matches the screen acquisition time with the sampling time of each reading item by item, and aggregates the screen, temperature reading, flow meter reading, electric valve opening reading, and sampling time record within the same sampling time period into a related data group, resulting in a time-period related data group; as detailed below: Based on the sampled information set, retrieve the pre-set sampling time period. for The millisecond, in physical terms, represents a typical mechanical response cycle of a fluid control system, and is defined as the current frame capture moment. ; According to the formula The calculated current target capture window range is: Milliseconds, retrieve all sensor sampling time records falling within the closed interval from the sampling information set, and set a given sampling time as... Millisecond spherical shroud drainage temperature reading Celsius, sampling time Millisecond flow meter reading cubic meters per hour and the corresponding electric valve opening Extract the data and compare the sampling times of each reading with... The logical relationship is to ensure that all time deviations are less than or equal to... Millisecond temperature readings, flow rate readings, valve status, and the frame image are physically clustered. For each group of clustered data, a sampling time is recorded as a unique index key. If multiple similar temperature readings are detected within a certain sampling period, the readings within that period are averaged to filter out high-frequency random thermal noise. This filtering process is predicated on the system's sampling frequency strictly meeting the Nyquist sampling criterion of being greater than or equal to twice the signal bandwidth. and Then take As the representative value of this associated data group, if no valid sensor reading is detected within the window, the time period is marked as a blank alarm state and the parameters of the previous valid window are used. The processed single frame and its corresponding environmental parameter set are encapsulated into an independent associated data unit to obtain the time period associated data group.

[0013] The air curtain writing submodule, based on the time-related data groups, calls the air compressor's air output status and the annular air outlet pipe's ejection status, marks the air curtain ejection duration interval along the continuous sampling time, and removes the related data groups whose image acquisition time falls within the air curtain ejection duration interval and whose corresponding airflow ejection status is not specified. The retained related data groups are written into the same time-series data sequence according to the sampling time order to obtain a synchronous record of the temperature image; the details are as follows: Based on the time-related data group, retrieve the high and low level status feedback from the air compressor control circuit (via the digital input DI port of the programmable logic controller PLC), and define the high level as the air output status. Low level indicates a silent state. Simultaneously, the value of the piezoresistive pressure sensor at the end of the annular outlet pipe is retrieved. If the pressure value exceeds the preset reference value... A pressure of megapascals (the minimum effective dust removal jet pressure required to overcome external atmospheric pressure) is considered the ejection state, and the ejection state is retrieved on the timeline. Become The start time and by Become The end time The duration of the air curtain ejection was determined to be... For example, the interval is the first milliseconds to the Milliseconds, considering the lag effect of pressure drop in the pipeline, based on the pipeline pressure decay time constant (in For flow resistance, A physical model is established for the capacitive reactance, if the voltage level has changed. However, the pipeline pressure is still higher than The safety threshold of megapascals will be Extending the timeline to the moment when the pressure is completely released, compare the acquisition time of each frame in the associated data group for each time period. If the acquisition time of a frame in a certain associated data group is... milliseconds and at this time the air compressor status bit is If the data set is found to be affected by air curtain occlusion, it is determined that the data set is interfered with and removed from the current sequence. For image data located in the penumbra at the edge of the air curtain, it is not directly and rigidly removed, but a Gaussian blur mask is introduced to give it a gradual removal weight to smooth the interference transition, retaining the data acquired at the specified time. milliseconds or Related data groups located outside the interval, such as milliseconds, will be rearranged according to the order of their sampling times after filtering. The synchronized data array will contain the pixel matrix of the image. Synchronization time Temperature vector at each point Flow scalar and valve opening These discrete sets of related data are written into a contiguous storage space to obtain a synchronized recording of temperature images.

[0014] The flame concentration identification module, based on the continuous images synchronously arranged in the sampling time order in the temperature image synchronous recording and the image acquisition time, calls the red, green, and blue channel readings and brightness readings of each pixel in the continuous images. It compares the red channel readings with the green channel readings point by point to identify red-green bias areas, and compares the red channel readings with the blue channel readings point by point to identify red-blue bias areas. The red-green bias areas and the red-blue bias areas are merged into red bias areas. Using the overall brightness state of the corresponding image as a reference, the brightness readings of each pixel are compared point by point to identify brightness offset areas. Pixels that are simultaneously in the red bias area and the brightness offset area are marked as candidate flame pixels. The candidate flame pixels are merged according to the spatial adjacency relationship to form a continuous flame area in a single frame. The continuous flame areas in the continuous images that are in the corresponding positions of adjacent images are compared frame by frame according to the image acquisition time to extract the continuously appearing areas with extended outer edges, thus obtaining the flame concentration area. The concentrated area of ​​flame includes the flame coverage, flame edge shape, flame extension shape, flame duration distribution, and flame brightness distribution.

[0015] Please see Figure 3 Specifically, the flame centralized identification module includes: The channel comparison submodule, based on the continuous images synchronously arranged in the sampling time sequence in the temperature image synchronous recording and the image acquisition time, calls the red, green, and blue channel readings of each pixel in the continuous images. It then compares the red channel readings with the green and blue channel readings point by point, merging the red-green and red-blue biased areas according to pixel position to obtain the red biased area; specifically as follows: The processor retrieves continuous frames and acquisition times from the temperature display synchronous recording, and iterates through the coordinates within the current frame. Extract the red channel reading for each of all pixels. Green channel readings and blue channel readings Set the red-green reference value for Red and blue color matching baseline value for The criterion value is based on the red zone of the flame in the CIE 1931 chromaticity diagram. And the flame spectral characteristics settings in the ASTM E135 standard, for a specific coordinate point in the image, such as... Read the value at that point; if that point for , for , for Perform a subtraction operation to calculate the difference between red and green values. get ,Will and To perform numerical comparisons, because Greater than The pixel is determined to belong to the red-green bias region; Similarly, perform subtraction to calculate the difference between red and blue. get ,Will and To perform numerical comparisons, because Greater than If a pixel is determined to belong to the red-blue bias region, then... Less than or equal to or Less than or equal to If the point does not possess significant red spectral characteristics, it is treated as background noise and removed, assigned a logical state of '0'. The entire pixel matrix is ​​traversed, and a corresponding binary marker matrix is ​​generated based on the judgment result. The pixel position information where the logical judgment result is true (i.e., simultaneously located in the red-green or red-blue biased region) is subjected to a union operation. For red flame pixels in the fire scene that are significantly affected by high-temperature radiation, their... Channel readings are typically within the range ,and and The channel remains in place due to its filtering properties. Within the interval, the coordinates of pixels that meet the conditions are spatially merged through the aforementioned point-by-point comparison and union processing to obtain the red biased region.

[0016] The brightness marking submodule calls the brightness readings of each pixel in the red bias area and the corresponding overall brightness state of the image. It compares the brightness readings of each pixel with the overall brightness state of the image point by point, marks the brightness offset area, and selects pixels that are simultaneously located in the red bias area and the brightness offset area to obtain candidate flame pixels; as detailed below: The brightness readings of the red-biased area and each pixel are retrieved, and the brightness characteristics of the current image are obtained through a weighted average method. For coordinates... For each pixel, call the standard grayscale conversion formula: Obtain the grayscale brightness readings, and calculate the arithmetic mean of the brightness readings of all pixels in the entire image as the overall brightness state of the image. For example, the average brightness of the current screen is Set the brightness offset judgment coefficient for (This coefficient is derived from the chi-square of actual fire scene measurement data) The statistical distribution of the test is used to distinguish between localized bright fire sources and global environmental reflections. To avoid misjudgment caused by weak background noise in dark environments, a setting is made. The effective lower limit is If the calculated Less than Then force it to be assigned Through formula The calculated brightness offset threshold is: Retrieve coordinates of points within the red-biased area The brightness reading is ,Will and To compare the amplitudes, because greater than If the brightness value of another reddish pixel is within the specified range, then that pixel is marked as a member of the brightness offset region. to If the comparison reveals that the value is within the range less than or equal to the threshold, it is determined to be a normal-temperature object near the fire scene affected by weak reflection. Its red bias marker is then removed. The aforementioned brightness determination action is then performed on each pixel defined within the red bias area. If the red bias marker at a certain coordinate point is... And the brightness offset flag is Then, the coordinate point is extracted using a logical AND gate operation, excluding background elements that, although reddish, have a brightness within a certain range. Interference from low-brightness, heat-generating objects in the region will be used to gather all pixels that pass the logical verification into a set of candidate pixel coordinates, thus obtaining candidate flame pixels.

[0017] The neighboring frame extraction submodule, based on candidate flame pixels, merges the pixel set within the same frame according to spatial adjacency, forming a continuous flame region within a single frame. It then calls the frame capture time and compares the corresponding continuous flame regions in adjacent frames frame by frame, filtering out continuously occurring regions with extended outer edges to obtain concentrated flame regions; specifically as follows: Based on the candidate flame pixels and their spatial adjacency, the connectivity of the candidate pixels in the horizontal, vertical, and diagonal directions is detected using the eight-neighbor connected component search algorithm (CCL technique). If the coordinates... and its adjacent coordinates, such as If all pixels exist in the candidate pixel set, then these spatially continuous pixels are marked with the same object identifier, forming a continuous flame region within a single frame. The image capture time is... The first frame region and its centroid coordinates and the second frame region sampled at adjacent time points. and its centroid coordinates Comparison and The degree of overlap in the image matrix is ​​used to set the outer edge extension coefficient. for By calculating the area ratio of the region Get as The dimensionless value will and Perform interval determination, because greater than Within the extended range, and determined The boundary pixel coordinate range exceeds If the original boundary range is determined, then the region is confirmed to satisfy the physical characteristics of continuous occurrence and outward expansion in temporal sequence. If the calculated ratio is... Less than or equal to This indicates that the bright area has not spread. At this point, it is necessary to supplement the measurement with an area shrinkage rate test. If the ratio... Less than If the area is shrinking, consistent with the characteristics of a fire attenuation phase or being extinguished, it should be marked as a heat source of attenuation for special tracking. Otherwise, it is highly likely to be a stationary high-temperature light fixture or a flickering reflective point, in which case subsequent tracking of that area should be immediately terminated. Frame to The frame repeatedly performs the aforementioned action of comparing the area growth slope with the center of gravity displacement, eliminating isolated light spots that flash momentarily in the fire scene, and retaining the set of pixels with a dynamic diffusion trend in spatial coordinates to obtain the concentrated area of ​​the flame.

[0018] The flue gas approach recognition module, based on the continuous images synchronously arranged in the sampling time sequence in the temperature image synchronous recording, calls up the grayscale readings, edge pixel counts, and low-contrast pixel proportions of each local area in the continuous images. Combined with the cone-shaped airflow coverage and airflow ejection direction formed by the circumferential dust removal air curtain device, it compares adjacent images frame by frame according to the image acquisition time. It identifies grayscale decrease areas by the changes in grayscale readings of each local area, edge weakening areas by the changes in edge pixel counts, and blurred diffusion areas by the changes in low-contrast pixel proportions. The grayscale decrease areas, edge weakening areas, and blurred diffusion areas are grouped into smoke climate selection areas according to the image position. Combined with the air curtain status, the cone-shaped airflow coverage and airflow ejection direction are mapped to the corresponding positions in the continuous images. Areas in the smoke climate selection areas that are within the cone-shaped airflow coverage and correspond to the airflow ejection direction are screened out. The position of the screened smoke climate selection areas is moved and compared according to the image acquisition time to determine the flue gas movement direction. The flue gas movement direction is compared with the approach direction from the smoke climate selection area to the flame concentration area to identify areas with the same direction, thus obtaining the flue gas obstruction approach area. The smoke shielding approach area includes the smoke diffusion range, smoke shielding boundary, smoke movement direction, smoke concentration distribution, and smoke approach area markers.

[0019] Please see Figure 4 Specifically, the flue gas approach detection module includes: The smoke extraction submodule, based on continuous images synchronously recorded in the temperature frame sequence and arranged according to sampling time, retrieves grayscale readings, edge pixel counts, and low-contrast pixel ratios for each local area. It then compares adjacent frames frame by frame according to the frame acquisition time to identify areas of decreasing grayscale, weakened edges, and blurred areas. These areas are then grouped by frame location to obtain the selected smoke climate region. Specifically: Based on the continuous frames synchronously arranged in the sampling time sequence from the temperature image synchronous recording, the processor performs grid-based segmentation on the first frame and extracts each... Grayscale readings of a local area of ​​a pixel If the gray value of a certain local area in the current sample is In adjacent samples, the gray values ​​are The grayscale change is obtained by performing the subtraction operation. Set the grayscale decrease judgment benchmark value for ,because Greater than The local area is marked as a region of decreasing grayscale, and the Sobel edge detection operator is simultaneously applied to this region, counting gradient magnitudes that are greater than a preset gradient threshold. The total number of pixels is used as the number of edge pixels. This gradient threshold value must be greater than the image resolution and the image signal-to-noise ratio. dB matching, in engineering terms, can also be equivalent to the gradient magnitude as a percentage of the total pixel value range. If the initial number of edges is The number of adjacent samples decreased to [number]. One, perform difference ratio calculation get Set the edge weakening threshold coefficient for ,because Greater than The region was determined to be an area with weakened edges, and the contrast value within this region was read to be lower than [value missing]. The proportion of low-contrast pixels is defined as the percentage of pixels in a region relative to the total number of pixels in that region. If this proportion changes from the initial... Increase to adjacent samples The execution logic compares the current percentage with the fuzzy benchmark value. The size relationship, because greater than The high proportion of the area is marked as the blur diffusion area. If any of the three conditions of gray level decrease, edge weakening or blur diffusion is not met, the change in the area is determined to be caused by ordinary moving shadow and is not regarded as smoke. The pixel coordinates that simultaneously satisfy the three attributes of gray level decrease, edge weakening and blur diffusion are set and merged to accurately locate the image blocks affected by smoke in the dynamic background of the fire. A two-dimensional array containing these discrete blocks is established by spatial coordinate mapping to obtain the smoke climate selection area.

[0020] The airflow mapping submodule, based on the smoke climate selection area, calls upon the cone-shaped airflow coverage area, airflow direction, and air curtain status formed by the circumferential dust removal air curtain equipment. It maps the cone-shaped airflow coverage area and airflow direction to continuous screen positions, filters each area within the smoke climate selection area, and removes areas located within the cone-shaped airflow coverage area and corresponding to the airflow direction, thus obtaining the jet stream exclusion area; specifically as follows: Based on the selected smoke climate area, the processor retrieves the preset relative position parameters between the PTZ camera's optical center and the circumferential dust removal air curtain outlet, and defines the projection equation of the cone-shaped airflow coverage area in the panoramic view. The specific physical parameters of this projection equation are set as the jet cone angle. Initial velocity of the jet meters per second, based on the current air compressor output status. and pressure readings Megapascals (MPa) confirms the air curtain is in the ejection state; retrieve the airflow ejection direction vector. In the current In a continuous frame of pixel resolution, the set of coordinates occupied by the conical airflow. Spatial overlay is performed, and the center coordinates of a certain block in the selected climate region are... Check if the coordinates fall within Within the covered sequence of coordinate points, perform a logical judgment if the coordinates belong to... Since the air curtain state is active at this moment, the displacement vector of this block between adjacent frames is extracted using the Lucas-Kanade optical flow method. Combine it with the direction vector of the airflow ejection. Substitute into the dot product formula Calculate the dimensionless cosine of the included angle and set a directional consistency threshold. for If the calculated cosine value is If the motion characteristics of the block are determined to be entirely caused by the air curtain jet, a removal action is performed to delete the coordinate block from the array of smoke climate selection areas. If the calculated cosine value is less than or equal to If the smoke flow direction is not significantly changed by the air curtain, it is retained. Only those independent smoke blocks located outside the coverage area of ​​the air curtain or whose motion vector does not match the airflow direction are retained. For each retained block, its vertex coordinates, center point coordinates, and area value are recorded to ensure that the subsequent analysis object is only the smoke flow direction generated by the actual fire scene, thus obtaining the jet exhaust area.

[0021] The direction determination submodule, based on the jet stream exclusion area, retrieves the image acquisition time and the flame concentration area, moves and compares each area to its corresponding position in adjacent images to determine the direction of smoke movement. It then compares this direction with the approaching direction from the selected smoke area to the flame concentration area, selecting areas with the same direction to obtain the smoke obstruction approaching area; specifically as follows: Based on the jet exhaust area, retrieve the image capture time and centroid coordinates of the flame concentration area obtained in the previous steps. In adjacent frame sequences, a center-of-gravity tracking operation is performed on each smoke block within the jet exhaust area, recording the center of gravity of the smoke block in the initial frame. centroid of adjacent frames The direction of flue gas movement is determined by performing vector subtraction. Simultaneously calculate the approach direction vector from the current flue gas centroid to the centroid of the concentrated flame region. Set the approach angle determination coefficient for Through formula Calculate the cosine similarity between the two, and substitute it into the actual example. If the flue gas movement vector is... The approach vector is Calculate the numerator dot product as ; The product of the denominator and the modulus is The above vector dot product operator conforms to the definition of Euclidean space and calculates the result. Approximately Compare this value with Comparison, because In the range Within the area, it is determined that the smoke mass is approaching the center of the flame and exhibiting a tendency to obstruct it; this area is marked as having the same direction. If the calculated... Less than or equal to If the smoke is moving away from the fire source or undergoing tangential movement and does not pose a blocking threat, it will not be marked. All candidate blocks within the jet exclusion area will be traversed, and all blocks with a cosine similarity greater than [value missing] will be [marked]. The blocks are logically grouped, and the specific pixel span and coverage area of ​​these blocks in the image are recorded to obtain the smoke obstruction approach area.

[0022] The heat zone extraction module, based on the continuously appearing and outer-edge-extending areas in the concentrated flame area and the directionally consistent areas in the smoke-shielding approaching area, calls the surface temperature readings of the transparent spherical cover synchronously recorded in the temperature screen. It aggregates the continuously appearing and outer-edge-extending areas in the concentrated flame area at corresponding positions in the continuous screen, identifies the concentrated flame distribution area, extends and compares the concentrated flame distribution area along the vertical direction of the screen to identify the longitudinally extended area, expands and compares the concentrated flame distribution area along the horizontal direction of the screen to identify the laterally expanded area, and matches the surface temperature readings of the transparent spherical cover with the concentrated flame distribution area, the longitudinally extended area, and the laterally expanded area to identify the temperature concentration area. It compares the temperature concentration area with the directionally consistent areas in the smoke-shielding approaching area, excludes the temperature concentration areas falling into the directionally consistent areas, and retains the temperature concentration areas corresponding to the concentrated flame area to obtain the high-heat concentration area. The high heat concentration area includes the heat accumulation range, the longitudinal heat expansion area, the lateral heat expansion area, the temperature concentration distribution, and the heat zone boundary information.

[0023] Please see Figure 5 Specifically, the hotspot extraction module includes: The fire zone aggregation submodule, based on continuously occurring and extending areas within the concentrated flame region, and areas with consistent direction in the approaching smoke obstruction region, calls the corresponding positions from the synchronously recorded temperature images. It aggregates these continuously occurring and extending areas according to their image positions, filters out contiguous areas, and obtains the concentrated flame distribution regions; specifically as follows: Based on the continuous and extended regions within the concentrated flame area, and the directionally consistent regions approaching the smoke obscuring area, the processor retrieves the coordinate indices of multiple consecutive frames from memory to extract the coordinates of each pixel within the previously identified flame outer edge extension region. Simultaneously extract the pixel coordinates of the region with consistent orientation within the smoke obstruction area. The DBSCAN density clustering algorithm is used, and the spatial neighborhood determination step size is set. for Pixels (in) The actual physical scale corresponding to the monitoring distance of meters is approximately (cm, consistent with the characteristics of an initial fireball), targeting a specific edge pixel within the flame region. Perform Euclidean distance calculation to retrieve its location. If another flame pixel is found within the range of all candidate pixels, then... ; By calculating the distance value Pixels, will and To perform a size comparison, because In less than Within the connected interval, the two points are grouped into the same connected set, and the red channel readings of the corresponding positions in the continuous images are read synchronously. Set the flame intensity benchmark value for If the average red channel reading of pixels within a contiguous set is ,Will and Numerical comparison was performed because greater than The significant flame region confirms that this connected region is the true flame core. If the distance between two points is greater than [missing information], then [missing information]. Pixels are then assigned to different clusters. For isolated pixels that cannot be assigned to any connected component, the algorithm assigns them the label of noise point. ', directly used as background noise for filtering, if the average red channel reading of the connected region is less than or equal to These are identified as residual heat reflection zones and removed. For multiple discretely distributed extended blocks in the image, a recursive search algorithm is executed to find all blocks that satisfy the condition of a spatial distance less than [a certain value]. Merge clusters of pixels that are consecutive in time sequence, and record the coordinates of the top-left corner of the smallest bounding rectangle of the merged region. and the coordinates of the bottom right corner This yields the concentrated area of ​​flame distribution.

[0024] The extended positioning submodule retrieves the surface temperature readings of the transparent dome of the spherical enclosure, which are synchronously recorded in the flame distribution concentration area and temperature image. It then extends and compares the area boundary along the vertical direction of the image, and expands and compares the area boundary along the horizontal direction of the image. The temperature readings are then correlated with the extended and expanded areas according to their positions to obtain the temperature concentration area; specifically as follows: The processor retrieves the surface temperature readings of the transparent dome of the spherical enclosure from the synchronously recorded temperature images of the concentrated flame distribution area. It then extracts the two-dimensional heat distribution matrix fed back from the infrared thermal imager or distributed patch sensor array on the surface of the transparent dome. To eliminate high-frequency spatial thermal noise from the sensor array, the sensor array is subjected to [a process / measurement] before extraction. implement The window uses median filtering to smooth the temperature gradient, retrieves the boundary coordinates of the concentrated flame distribution area, and moves it upwards along the vertical axis of the screen. Decrease direction and downward Increase direction of expansion Pixels, as a vertical extension, extend to the left along the horizontal axis of the image. Decrease direction and right side Increase direction of expansion Pixels are used as the lateral expansion factor, and a temperature correlation coefficient is set. for By reading coordinate points The surface temperature reading at that location is Celsius, performing bilinear interpolation to accurately map discrete physical temperature readings to a two-dimensional pixel coordinate system, setting a high-temperature judgment benchmark value. for Celsius, the temperature reading at the current coordinate point. and Perform numerical judgment, because Greater than The coordinate point is marked as a thermal correlation point. For pixels within the vertically extended region, the aforementioned temperature matching action is performed. If the coordinate point above is detected... The temperature reading at is If the temperature is in degrees Celsius, then that point is included in the temperature coverage range; if the interpolated temperature at a point is lower than or equal to... If the temperature is below a certain degree Celsius, it indicates that the pixel was not substantially affected by the thermal convection of the fire, and it is excluded from the temperature coverage area. This is achieved by iterating through and comparing the sensor sampling values ​​corresponding to each pixel location within the extended area, filtering out all temperature readings within this range. The set of coordinates within a Celsius range is used to lock the boundaries of connected pixels that satisfy the positional correspondence and whose values ​​reach the high-temperature benchmark, thus obtaining the temperature concentration region.

[0025] The smoke masking and exclusion submodule, based on the temperature concentration area, calls the directionally aligned areas and flame concentration areas within the smoke masking approach area. It then filters these temperature concentration areas and directionally aligned areas according to their position on the screen, excluding those falling within the directionally aligned areas and retaining the temperature concentration areas corresponding to the flame concentration areas, thus obtaining the high-heat concentration areas; specifically as follows: Based on the temperature concentration region, the processor retrieves the coordinate mask set of the temperature concentration region by calling the directionally consistent region and the flame concentration region in the smoke obstruction approach region. Synchronously retrieve the coordinate mask set of the region with the same direction. ,for any coordinate point in Perform a logical retrieval operation to determine whether the point coincides with... Within the set, an overlap determination coefficient is set. for This means that when the smoke coverage area exceeds Visual feature failure is immediately determined by calculating the ratio of the intersection area of ​​the two regions to the total area of ​​the temperature concentration region. Substituting into the actual calculation example, if the area of ​​the temperature region is... Pixels, where the overlapping area of ​​the region approaching the smoke is Pixels, obtained by performing division. ,Will and Compare sizes, because Greater than The temperature concentration area is determined to be mainly caused by high-temperature flue gas flow causing temperature rise or by flue gas blocking infrared radiation, resulting in visual artifacts. Therefore, a rejection instruction is executed to remove this area from the candidate sequence. If the area of ​​another temperature region is... Pixels and only The pixel overlaps with the smoke, and their ratio is In less than If the non-interference region is not specified, a retention action is performed, and the retained region is compared with the centroid of the preceding flame concentration region. Perform distance verification; if the center point displacement deviation is less than... If the pixel value is large enough, then the region is determined to be the high-temperature core area directly radiated by the flame. If the center displacement deviation is greater than or equal to... For pixels, the area is downgraded to a secondary heat source and not given core attention. All temperature coordinates that pass the logical filtering and are precisely matched with the flame position are clustered and stored to obtain the high heat concentration area.

[0026] The flow direction switching module calls the temperature readings of each circumferential position of the annular cooling channel synchronously recorded on the temperature screen, as well as the temperature concentration area corresponding to the flame concentration area retained in the high heat concentration area; according to the sampling time sequence, the temperature readings of the same circumferential position at adjacent sampling times are continuously compared to identify the circumferential position of continuous heating; according to the circumferential orientation of the spherical cover, the circumferential position of continuous heating is matched with the position of the temperature concentration area; the circumferential position of heat concentration that simultaneously meets the requirements of continuous heating and the corresponding high heat concentration area is identified; the circumferential position of heat concentration is compared with the preset flow channel starting direction to determine the cooling medium flow direction switching direction; according to the cooling medium flow direction switching direction, the opening and closing states of the inlet water channel control valve and the return water channel control valve are adjusted to obtain the panoramic monitoring execution results; The panoramic monitoring results include flow direction switching status, valve control status, cooling medium flow status, heat concentration location, and cooling adjustment results.

[0027] Please see Figure 6 Specifically, the flow switching module includes: The temperature rise identification submodule calls the temperature readings of each circumferential position of the annular cooling channel synchronously recorded in the temperature screen, as well as the temperature concentration area corresponding to the flame concentration area retained in the high heat concentration area. According to the sampling time sequence, the temperature readings of the same circumferential position at adjacent sampling times are continuously compared to identify the circumferential position of continuous temperature rise and obtain the temperature rise circumferential position. The process of continuously comparing temperature readings at the same circumferential location at adjacent sampling times to identify the circumferential location of continuous temperature rise is as follows: Temperature readings at the same circumferential location during continuous sampling are arranged chronologically. The trends in temperature readings at adjacent sampling times are compared sequentially. When the temperature readings consistently show an upward trend during continuous sampling, the location is identified as a continuously heating circumferential location. Specifically: The processor retrieves temperature readings and high-heat concentration areas from the annular cooling channel, which are synchronously recorded on the temperature screen, and extracts the data from the memory marked as... , , and Raw data sequences from four PT100 sensors, with the continuous sampling window length set. for (i.e., includes) Each sampling time interval If the single sampling period is Seconds, extract historical temperature values ​​from the current sample to the four previous samples, targeting The position is subtracted to obtain the change between adjacent time points. If the sequence value is read , , , and The Celsius value is then calculated sequentially to obtain four consecutive increments. , , , Celsius, set the temperature rise judgment threshold for The threshold value is set in degrees Celsius, based on the premise that the minimum measurable temperature rise is approximately equal to the sensor accuracy multiplied by [a certain value]. Right now Temperature adjustment can effectively eliminate system background noise and compare each incremental value with... Perform a numerical comparison; if all increments are greater than... If the temperature rises in the positive growth range, then the counting logic is executed to accumulate the number of temperature rises. When the count result equals... Right now At this time, it is determined that the specific circumferential orientation is in an abnormal heat conduction state, and the center coordinates of the high heat concentration area are retrieved. If any of the following occurs during this judgment process Less than or equal to Negative values ​​in Celsius indicate inconsistent temperature changes in the monitoring area, suggesting interference from alternating hot and cold air currents or sensor malfunction. The processor immediately resets the temperature rise counter and restarts the next round of timing checks. If the temperature rise condition is met, the coordinates are mapped to the preset annular flow channel spatial coordinate system for calculation. The angular distance between the physical coordinates of the position sensor and the coordinates of the high-temperature center, if the angular distance is less than... And the average temperature rise rate obtained using the formula: Celsius per second, due to If the fire exhibits significant heat absorption characteristics and conforms to the fire power growth model in the NFPA 92 standard, then it is confirmed that the location is significantly affected by the heat source radiation of the fire. By repeatedly performing the above time-domain slope comparison and spatial coordinate mapping on all flow channel locations, the location identifier that conforms to the heating logic is locked, and the circumferential heating location is obtained.

[0028] The orientation correspondence submodule, based on the circumferential position of the heating, calls the circumferential orientation of the spherical cover and the temperature concentration area retained in the high heat concentration area, and matches the circumferential position of the continuous heating with the position of the temperature concentration area according to the circumferential orientation, and filters the circumferential positions that simultaneously correspond to the continuous heating and the high heat concentration area to obtain the circumferential position of heat concentration. The process of mapping the location of the continuously rising circumferential position to the location of the temperature concentration area according to the circumferential orientation is as follows: A correspondence is established between the circumferential orientation of the spherical cover and its spatial position in continuous images. The circumferential position of continuous heating is mapped to the corresponding image orientation area, and the image position of the temperature concentration area is converted into the corresponding circumferential orientation. When the circumferential position of continuous heating and the temperature concentration area coincide in circumferential orientation, it is determined as the circumferential position of heat concentration; the details are as follows: Based on the circumferential position of the heating element, the processor retrieves the intrinsic parameter calibration matrix of the panoramic PTZ camera calibrated using a pinhole camera model, and employs the Brown-Conrady model to correct radial and tangential distortion of the lens, establishing pixel coordinates in the continuous images. With the circumferential azimuth of the dome The mapping function sets the center vertical axis of the image as the circumferential axis. Based on the reference benchmark, for the high-heat concentrated areas identified in the image, extract the set of all pixel coordinates covered by them. Calculate the average level component of pixels within the set. Substituting the values ​​into the mapping equation, the circumferential azimuth angle corresponding to the high-heat concentration region is calculated. for Simultaneously extract the circumferential heating position determined in the previous steps, such as The physical range corresponding to the sensor Execute the judgment logic check of the interval. If the location falls within the specified physical range, and the result is true, then the average temperature rise rate for that location is extracted. Average brightness value of high heat area Weighted calculation after normalization: The above and The weighting coefficients are dynamically calculated using the information entropy weighting method to ensure the objective fusion of multi-source data; Calculated The dimensionless heat concentration evaluation index is obtained as follows: Set a baseline value for heat consistency verification. for ,Will and Compare sizes, because greater than The high matching range confirms that the characteristics of high-heat images and the physical temperature rise of the flow channel are homologous in spatial dimension. Not falling within the physical range or the calculated index is less than or equal to This indicates that the discrepancy between the visual image and the physical temperature rise may be caused by specular reflection. Immediately block the signal association in that direction and then check the spherical cover in its entirety. The topological overlap between visual hotspots and sensor hotspots within the range is used to filter out visual false hotspots caused by reflections from external light sources, accurately pinpoint the flow channel sector aligned with the real heat source, and obtain the circumferential location of heat concentration.

[0029] The valve switching submodule, based on the circumferential location of heat concentration, calls the preset flow channel starting direction, water inlet channel control valve, and water return channel control valve. It compares the circumferential location of heat concentration with the preset flow channel starting direction to determine the cooling medium flow direction switching direction, and adjusts the valve opening and closing states according to the cooling medium flow direction switching direction to obtain the panoramic monitoring execution results. The process of comparing the circumferential location of heat concentration with the starting direction of the preset flow channel is as follows: Using the preset flow channel starting direction as a reference direction, the circumferential distribution of the heat concentration position relative to the preset flow channel starting direction is determined. Based on the relative positional relationship of the heat concentration position relative to the preset flow channel starting direction, the cooling medium flow direction switching direction is determined, and the opening and closing states of the inlet water channel control valve and the return water channel control valve are adjusted accordingly according to the cooling medium flow direction switching direction; specifically as follows: Based on the circumferential location of heat concentration, the processor retrieves the current status information of the inlet channel control valve (servo motor-driven proportional regulating ball valve) and the return channel control valve, and reads that the preset flow direction is clockwise and the starting inlet point is located at... Orientation, referring to the circumferential location of the identified heat concentration. Calculate its radian displacement relative to the initial direction. Set the shortest path cooling determination logic, and compare the clockwise path distance. Distance from counterclockwise path In the current example for and for Perform numerical comparison and judgment Within a smaller value range, the cooling medium flow direction is determined to be clockwise, thus enhancing the flow. If a special location occurs that... and Completely equal (e.g., the target heat source is located at) If the flow distribution mechanism is activated (at a certain point), it will automatically trigger the flow equalization mechanism, opening the valves on both sides of the flow channel by the same proportion. Before the flow equalization mechanism is executed, a hydraulic resistance model based on the Darcy-Weisbach formula must be substituted, and flow conservation verification must be performed to ensure the total influent standard volumetric flow rate of the system. Equal to the total standard volumetric flow rate of the outflow retrieve the electric valve opening reading Perform incremental control calculations to adjust the location corresponding to the heat concentration point. The target opening degree of the branch inlet valve is adjusted to... At the same time, it will be in a reverse position Nearby The target opening degree of the branch inlet valve is adjusted to... The logic controller sends a pulse width modulation (PWM) signal to the control terminal to change the speed of the DC motor, thus executing the physical action. The duty cycle calculation of the PWM signal must be strictly correlated with the time constant of the actuator to prevent overshoot or oscillation. The readings of the electromagnetic flowmeter are monitored synchronously. Standard cubic meters per hour to target value The standard cubic meter per hour shift is adjusted in real time according to the flow direction switching, changing the on / off combination of the electromagnetic reversing valve to make the cooling medium converge on the inner side of the flow channel corresponding to the high heat area to achieve targeted rapid cooling. The output includes the instruction execution log containing valve opening, flow rate change and orientation correspondence, and obtains the panoramic monitoring execution results.

[0030] The above description is merely a specific embodiment of the present invention, but the scope of protection of the present invention is not limited thereto. Any variations or substitutions that can be easily conceived by those skilled in the art within the technical scope disclosed in the present invention should be included within the scope of protection of the present invention. Therefore, the scope of protection of the present invention should be determined by the scope of the claims.

Claims

1. An active cooling fire scene panoramic monitoring system, characterized in that, The system includes: The timing alignment processing module acquires continuous images, image acquisition time, temperature readings, flow meter readings, cooling medium readings, electric valve readings, and air curtain status. It collects related data groups according to the sampling time corresponding to the preset time period, removes related data groups affected by airflow based on the air curtain ejection interval, and writes them into the same sequence to obtain synchronous recording of temperature images. The flame concentration identification module, based on the continuous images and acquisition time recorded synchronously in the temperature image, calls the pixel red, green and blue and brightness readings, compares the red and green, red and blue bias and brightness shift, marks the candidate flame pixels, merges them according to spatial adjacency and extracts the continuous outer edge extension area frame by frame to obtain the flame concentration area. The smoke approach recognition module, based on the temperature image synchronously recorded, calls grayscale, edge and low contrast ratio to identify the smoke area, combines the air curtain coverage range and the ejection direction to filter out airflow interference, and compares the area with the same moving direction and approaching direction at any time to obtain the smoke obstruction approach area. The heat zone extraction module, based on the flame concentration area and the smoke obstruction approach area, calls the surface temperature reading of the transparent cover, collects the flame distribution and identifies the longitudinal and lateral expansion areas, corresponding to the temperature concentration areas, and excludes the parts that fall into the area with the same direction, thus obtaining the high heat concentration area.

2. The active cooling fire scene panoramic monitoring system according to claim 1, characterized in that: The temperature display synchronously records information including display time correspondence, temperature correlation, air curtain status, flow rate correspondence, valve status, and circumferential temperature distribution. The flame concentration area includes flame coverage, flame edge shape, flame extension shape, flame continuity distribution, and flame brightness distribution. The smoke obstruction approach area includes smoke diffusion range, smoke obstruction boundary, smoke movement direction, smoke concentration distribution, and smoke approach area marker. The high heat concentration area includes heat accumulation range, longitudinal heat expansion area, lateral heat expansion area, temperature concentration distribution, and heat zone boundary information.

3. The active cooling fire scene panoramic monitoring system according to claim 1, characterized in that: The timing alignment processing module includes: The information access submodule acquires continuous images captured by the panoramic camera PTZ camera, image acquisition time, PTZ shroud drainage temperature reading, PTZ shroud transparent cover surface temperature reading, PTZ circumferential position temperature reading of the annular cooling channel, cooling medium inlet temperature reading, cooling medium outlet temperature reading, flow meter reading, electric valve opening reading, and sampling time record corresponding to each reading. It organizes the correspondence between images and readings according to the sampling time record and generates a set of sampling information. The time period collection submodule, based on the sampling information set, calls the screen acquisition time, the sampling time record corresponding to each reading, and the preset sampling time period, and matches the screen acquisition time with the sampling time of each reading item by item. It collects the screen, temperature reading, flow meter reading, electric valve opening reading and sampling time record in the same sampling time period into a related data group to obtain the time period related data group. The air curtain writing submodule, based on the time period associated data group, calls the air compressor outlet status and the annular outlet pipe spray status, marks the air curtain spray duration interval along the continuous sampling time, removes the associated data group whose image acquisition time falls within the air curtain spray duration interval and whose corresponding airflow spray status is removed, and writes the retained associated data group into the same time sequence data sequence according to the sampling time order to obtain the temperature image synchronous record.

4. The active cooling fire scene panoramic monitoring system according to claim 1, characterized in that: The flame concentration identification module includes: The channel comparison submodule, based on the continuous images synchronously arranged in the sampling time order in the temperature image synchronous record and the image acquisition time, calls the red channel, green channel and blue channel readings of each pixel in the continuous image, compares the red channel readings with the green channel readings and blue channel readings point by point, and merges the red-green bias area and the red-blue bias area according to the pixel position to obtain the red bias area; The brightness marking submodule calls the red bias area, the brightness readings of each pixel in the continuous image, and the overall brightness state of the corresponding image. It judges the brightness readings of each pixel and the overall brightness state of the corresponding image point by point, marks the brightness offset area, and selects the pixels that are simultaneously located in the red bias area and the brightness offset area to obtain the candidate flame pixels. The neighboring frame extraction submodule, based on the candidate flame pixels, merges the pixel set within the same frame according to the spatial adjacency relationship to form a continuous flame region within a single frame. It calls the frame acquisition time and compares the corresponding continuous flame regions in adjacent frames frame by frame, filters out regions that appear continuously and have extended outer edges, and obtains the concentrated flame region.

5. The active cooling fire scene panoramic monitoring system according to claim 1, characterized in that: The flue gas proximity recognition module includes: The smoke extraction submodule, based on the continuous images synchronously arranged in the sampling time order in the temperature image synchronous recording, calls up the grayscale readings, edge pixel counts and low contrast pixel ratios of each local area image, compares adjacent images frame by frame according to the image acquisition time, identifies grayscale decrease areas, edge weakening areas and blur diffusion areas, and collects them according to the image position to obtain the smoke climate selection area. The airflow mapping submodule, based on the smoke climate selection area, calls the cone-shaped airflow coverage range, airflow ejection direction and air curtain status formed by the circumferential dust removal air curtain device, maps the cone-shaped airflow coverage range and airflow ejection direction to the continuous screen position, filters each area in the smoke climate selection area, removes the area located within the cone-shaped airflow coverage range and corresponding to the airflow ejection direction, and obtains the jet stream exclusion area. The direction determination submodule, based on the jet stream exclusion area, calls the image acquisition time and the flame concentration area, moves and compares each area in the corresponding position of adjacent images to determine the direction of smoke movement, compares the direction of smoke movement with the approach direction from the smoke selection area to the flame concentration area, selects the area with the same direction, and obtains the smoke obstruction approach area.

6. The active cooling fire scene panoramic monitoring system according to claim 1, characterized in that: The heat zone extraction module includes: The fire zone aggregation submodule, based on the continuously appearing and outer-edge extended areas in the concentrated flame area and the directional areas in the smoke-obstructed approach area, calls the corresponding positions of the continuous images in the temperature image synchronous recording, aggregates the continuously appearing and outer-edge extended areas according to the image position, filters the connected positions of the areas, and obtains the concentrated flame distribution area. The extended positioning submodule calls the temperature readings of the transparent spherical cover surface recorded synchronously in the flame distribution concentration area and the temperature screen, extends and compares the area boundary along the vertical direction of the screen, expands and compares the area boundary along the horizontal direction of the screen, and matches the temperature readings with the extended and expanded areas according to their positions to obtain the temperature concentration area. The smoke masking and exclusion submodule, based on the temperature concentration area, calls the directional consistent area and flame concentration area in the smoke masking approach area, filters the temperature concentration area and directional consistent area according to the screen position, excludes the temperature concentration area falling into the directional consistent area, and retains the temperature concentration area corresponding to the flame concentration area to obtain the high heat concentration area.

7. The active cooling fire scene panoramic monitoring system according to claim 1, characterized in that: The system also includes: The flow direction switching module calls the temperature screen synchronous record and the high heat concentration area, continuously compares the temperature readings at adjacent sampling times at the same circumferential position, identifies the circumferential position that is continuously heating up and corresponds to high heat, compares it with the preset flow channel starting direction and adjusts the valve to obtain the panoramic monitoring execution result; The results of the panoramic monitoring include flow direction switching status, valve control status, cooling medium flow status, heat concentration location, and cooling adjustment results.

8. The active cooling fire scene panoramic monitoring system according to claim 7, characterized in that: The flow direction switching module includes: The temperature rise identification submodule calls the temperature readings of each circumferential position of the annular cooling channel synchronously recorded in the temperature screen, as well as the temperature concentration area corresponding to the flame concentration area retained in the high heat concentration area. According to the sampling time sequence, the temperature readings of the same circumferential position at adjacent sampling times are continuously compared to identify the circumferential position of continuous temperature rise and obtain the circumferential position of temperature rise. The orientation correspondence submodule, based on the heating circumferential position, calls the circumferential orientation of the spherical cover and the temperature concentration area retained in the high heat concentration area, and matches the continuous heating circumferential position with the position of the temperature concentration area according to the circumferential orientation, and filters the circumferential positions that simultaneously correspond to the continuous heating and the high heat concentration area to obtain the heat concentration circumferential position. The valve switching submodule, based on the circumferential position of the heat concentration, calls the preset flow channel starting direction, the inlet channel control valve, and the return channel control valve. It compares the circumferential position of the heat concentration with the preset flow channel starting direction to determine the cooling medium flow direction switching direction, and adjusts the valve opening and closing states according to the cooling medium flow direction switching direction to obtain the panoramic monitoring execution result.

9. The active cooling fire scene panoramic monitoring system according to claim 8, characterized in that: The process of continuously comparing temperature readings at the same circumferential location at adjacent sampling times to identify the circumferential location of continuous temperature rise is as follows: Arrange the temperature readings at the same circumferential position in chronological order during continuous sampling. Compare the temperature reading trends at adjacent sampling times. When the temperature readings show an upward trend during continuous sampling, it is determined to be a circumferential position with continuous temperature rise. The process of mapping the circumferential position of the continuous heating with the position of the temperature concentration area according to the circumferential orientation is as follows: A correspondence is established between the circumferential orientation of the sphere cover and the spatial position in the continuous image. The circumferential position of the continuous heating is mapped to the corresponding image orientation area, and the image position where the temperature concentration area is located is converted into the corresponding circumferential orientation. When the circumferential position of the continuous heating and the temperature concentration area coincide in the circumferential orientation, it is determined as the circumferential position of heat concentration.

10. The active cooling fire scene panoramic monitoring system according to claim 8, characterized in that: The process of comparing the circumferential location of heat concentration with the starting direction of the preset flow channel is as follows: Using the preset flow channel starting direction as a reference direction, the circumferential distribution of the heat concentration circumferential position relative to the preset flow channel starting direction is determined. Based on the relative positional relationship of the heat concentration circumferential position relative to the preset flow channel starting direction, the cooling medium flow direction switching direction is determined, and the opening and closing states of the inlet channel control valve and the return channel control valve are adjusted accordingly according to the cooling medium flow direction switching direction.

Images