Building fireproof detection data processing system and method based on multi-source data fusion

Abstract

The invention provides a building fireproof detection data processing system and method based on multi-source data fusion, and belongs to the technical field of building fireproof detection data processing, and the system comprises a source data collection module which collects dynamic video image data, fire-fighting sensing parameters and fire-fighting equipment state monitoring data in a building in real time; the data intelligent analysis module is used for performing feature extraction on the dynamic video image data by adopting an AI algorithm to obtain video features, and judging whether a fire occurs in the building or not to obtain a video analysis result; the multi-source data fusion and decision module is used for fusing the video analysis result, the fire-fighting sensing parameters and the fire-fighting equipment state monitoring data analysis result to obtain a final building fire safety detection result, and fusing video features to generate fire scene features; and the fire report sending module is used for generating a fire scene report based on the fire scene characteristics and the final building fire safety detection result so as to provide powerful guarantee for building fire safety.

Description

TECHNICAL FIELD

[0001] The present application relates to the technical field of building fire prevention detection data processing, and particularly relates to a building fire prevention detection data processing system and method based on multi-source data fusion. BACKGROUND

[0002] In the field of building fire safety management, the traditional detection methods have many limitations. On the one hand, the method of relying on manual regular inspection is not only inefficient, but also subjective and easy to miss some potential fire hazards. On the other hand, the existing automatic detection equipment, such as simple smoke sensors and temperature sensors, can only provide single-dimensional information and cannot comprehensively and systematically evaluate the building fire situation as a whole. For example, only detecting the increase of smoke concentration through the smoke sensor may not be able to accurately determine whether the smoke is generated in the early stage of fire or caused by dust rising due to cleaning operation. At the same time, the traditional detection methods are difficult to monitor the important fire safety factors such as whether the evacuation passage of the building is unobstructed and whether the fire-fighting facilities are blocked in real time and dynamically. With the continuous expansion of building scale and the increasing complexity of functions, higher requirements are put forward for the accuracy, comprehensiveness and intelligence of building fire detection. Therefore, a technical solution that can comprehensively utilize multiple data sources and realize omnibearing and multi-level detection through intelligent analysis is urgently needed. SUMMARY

[0003] The present application provides a building fire detection data processing system and method based on multi-source data fusion. By introducing a video monitoring intelligent analysis method, the system can systematically, completely, omnibearing and multi-levelly implement intelligent detection and analysis of building fire prevention, effectively make up for the shortcomings of traditional detection methods, improve the accuracy and reliability of building fire detection, and provide strong protection for building fire safety.

[0004] The present application provides a building fire detection data processing system based on multi-source data fusion, comprising: a multi-source data acquisition module for acquiring dynamic video image data, fire-fighting sensor parameters and fire-fighting equipment state monitoring data in real time; a data intelligent analysis module for extracting features from the dynamic video image data using AI algorithms to obtain video target features, on-site personnel behavior features and fire features, and determining whether a fire occurs in the building to obtain a video analysis result; wherein the video target includes personnel, fire-fighting facilities, obstacles and flames; The multi-source data fusion and decision module is configured to fuse the video analysis result with the fire-fighting sensing parameter and the fire-fighting equipment state monitoring data analysis result to obtain a final building fire safety detection result, and to fuse the video target feature, the on-site personnel behavior feature and the fire feature to generate a fire scene feature. The fire report sending module is configured to generate a fire scene report based on the fire scene feature and the final building fire safety detection result, and send the fire scene report to a building manager and a relevant fire department.

[0005] Preferably, in the building fire detection data processing system based on multi-source data fusion, the data intelligent analysis module comprises: The video processing unit is configured to perform real-time video frame segmentation on the dynamic video image data to obtain a real-time monitoring picture, and mark a plurality of video targets in the real-time monitoring picture by using different color rectangular frames based on an AI algorithm. The fire determination unit is configured to determine whether there is a fire marking frame in the current video target marking result, and if there is, determine that a fire occurs in the building, determine a fire occurrence position according to a real-time monitoring picture collection position, and generate a fire-fighting early warning signal and send the fire-fighting early warning signal to a monitoring management end. The video feature extraction unit is configured to extract a video target feature based on the video target marking result. When the video target is determined to be a person, a behavior feature of the person is extracted to obtain an on-site personnel behavior feature. When it is determined that there is a fire marking frame in the current video target marking result, a fire feature is extracted from a local picture in the fire marking frame.

[0006] Preferably, in the building fire detection data processing system based on multi-source data fusion, the data intelligent analysis module further comprises: The target state detection unit is configured to determine positions of each sub-building in a building effective area based on the video target marking result, position fire safety evacuation and fire rescue facilities in the building effective area in combination with a fire-fighting processing facility demand and building layout data, and determine existing fire safety evacuation and fire rescue facility positions and missing fire safety evacuation and fire rescue facilities in the building effective area according to the positioning result, and generate a fire-fighting facility missing report according to the missing fire safety evacuation and fire rescue facilities and send the fire-fighting facility missing report to relevant management personnel. According to the positioning result, determine existing fire safety evacuation and fire rescue facility positions and missing fire safety evacuation and fire rescue facilities in the building effective area, and generate a fire-fighting facility missing report according to the missing fire safety evacuation and fire rescue facilities and send the fire-fighting facility missing report to relevant management personnel. Meanwhile, the actual reserved position size is calculated based on an image scale of the real-time monitoring picture and an actual image size of the existing fire safety evacuation facility in the building, a fire-fighting facility report is generated and sent to an associated APP for display. When a fire occurs in the building, a fire-fighting guidance report is generated and sent to relevant management personnel.

[0007] Preferably, in a building fire protection detection data processing system based on multi-source data fusion, the data intelligent analysis module further includes: The duty monitoring unit is used to acquire and analyze video target features and on-site personnel behavior characteristics in the fire control room to determine whether the current number of people on duty in the fire control room is consistent with the prescribed number for the current time period. If there is a discrepancy, an anomaly notification shall be sent to the fire control room manager and the on-duty status of the fire control room shall be monitored. If they match, determine if there is any abnormal behavior by the on-duty personnel. If so, send a behavioral alert voice message to the fire control room. Otherwise, continue to monitor the duty situation in the fire control room.

[0008] Preferably, in a building fire protection detection data processing system based on multi-source data fusion, the multi-source data fusion and decision-making module includes: The decision fusion unit is used to determine whether the detection of each fire sensor is reliable based on the status monitoring data of fire equipment, fuse the analysis data results corresponding to reliable fire sensors with video analysis, determine the range of fire-affected area and the location of the fire point, and generate the final building fire safety detection result based on the range of fire-affected area and the location of the fire point. The on-site video feature fusion unit is used to, when it is determined that there is a fire in the building, take the location of the fire as the target location, and acquire all relevant dynamic video image data of the target location and its corresponding video target features, on-site personnel behavior features, and fire features; Based on the shooting angle information of relevant dynamic video images, the overlapping detection area is determined. Based on the shooting time and the overlapping detection area, the relevant dynamic video image data is stitched together to obtain a panoramic fire image. Based on the panoramic fire image, the video target features, on-site personnel behavior features, and fire features corresponding to all relevant dynamic video image data are fused to obtain panoramic video target features, panoramic on-site personnel behavior features, and panoramic fire features, respectively. The characteristics of the fire scene are obtained by fusing the panoramic video target features, panoramic on-site personnel behavior features, and panoramic fire features.

[0009] Preferably, in a building fire detection data processing system based on multi-source data fusion, the on-site video feature fusion unit includes: The first panoramic feature fusion subunit is used to obtain the first image proportion of each relevant dynamic video image data on the fire panoramic image, and to obtain the first fusion weight. Based on the first fusion weight, the video target features and on-site personnel behavior features of all relevant dynamic video images are fused to obtain panoramic video target features and panoramic on-site personnel behavior features. The second panoramic feature fusion subunit is used to obtain the second image ratio between the first fire center region image on the fire panoramic image and the second fire center region image corresponding to each relevant dynamic video image, and to obtain the second fusion weight. Based on the second fusion weight, the fire features of all relevant dynamic video images are fused to obtain panoramic fire features.

[0010] Preferably, in a building fire detection data processing system based on multi-source data fusion, the fire determination unit further includes: The fire analysis and guidance subunit is used to receive and analyze the fire scene characteristics fed back by the multi-source data fusion and decision-making module to determine the optimal fire handling route, including: The safety level determination subunit is used to determine the relative distance between each video target and the fire center area based on the characteristics of the panoramic video targets, and to determine the fire spread speed and fire size according to the fire characteristics. Based on the fire spread rate and fire size, and combined with the relative distance between different video targets and the fire center area, the current safety status of different video targets in the current fire is determined, and based on the current safety status, relatively safe areas and dangerous areas are determined on the panoramic fire image. The on-site evacuation guidance subunit is used to determine the actual movement direction of on-site personnel based on the characteristics of the fire scene, and to determine whether there are personnel moving towards the danger zone at the fire scene; If they exist, the personnel on site will be divided into multiple evacuation directions based on the location of the safety escape exits, and the name of the landmark item in each direction and its relative location data with the corresponding escape route will be determined. Based on the names of iconic objects and their relative locations to the corresponding escape routes, combined with the road division data within the building, voice broadcast information guiding the evacuation routes in the corresponding evacuation directions is generated and broadcast.

[0011] Preferably, in a building fire detection data processing system based on multi-source data fusion, the fire analysis and guidance subunit further includes: The intelligent rescue guidance subunit is used to identify on-site personnel moving towards the danger zone as rescue targets based on the characteristics of the fire scene, and to mark, track, and count them to obtain the real-time dynamic trajectory and number of rescue targets. The marking and tracking are removed after the rescue targets leave the danger zone. Based on the location of the boundary between the dangerous area and the relatively safe area, determine the available fire-fighting equipment nearby, and control the sound positioning alarm on the available fire-fighting equipment to emit a warning sound; At the same time, based on the characteristics of the fire scene, the weak points of the fire in the central area of ​​the fire are determined. Based on the weak points and the real-time dynamic trajectory of the target to be rescued, multiple recommended priority rescue routes are obtained. After marking the locations of all targets to be rescued on each priority rescue route, the information is sent to firefighters or building managers.

[0012] Preferably, in a building fire detection data processing system based on multi-source data fusion, the fire analysis and guidance subunit further includes: The target trajectory analysis subunit is used to obtain the actual running trajectory and moving speed of the target on the panoramic image of the fire scene when the video surveillance equipment at the fire scene is damaged and the tracking of the target fails. Based on the actual running trajectory and movement speed, the running trajectory of the target to be rescued is predicted, the predicted trajectory is obtained, and the predicted trajectory is sent to the intelligent rescue guidance subunit as a substitute trajectory for the real-time movement trajectory of the target to be rescued.

[0013] This invention provides a method for processing building fire detection data based on multi-source data fusion, including: Step 1: Real-time acquisition of dynamic video image data, fire sensor parameters, and fire equipment status monitoring data within the building; Step 2: Use AI algorithms to extract features from dynamic video image data to obtain video target features, on-site personnel behavior features, and fire features, determine whether a fire has occurred inside the building, and obtain video analysis results; The video targets include people, fire-fighting equipment, obstacles, and flames; Step 3: Integrate the video analysis results with the fire sensor parameters and the fire equipment status monitoring data analysis results to obtain the final building fire safety test results. At the same time, integrate the video target features, on-site personnel behavior features and fire features to generate fire scene features. Step 4: Based on the characteristics of the fire scene and the final building fire safety inspection results, generate a fire scene report and send it to building management personnel and relevant fire departments.

[0014] Compared with the prior art, the present invention has at least the following beneficial effects: This invention simultaneously collects dynamic video image data, fire sensor parameters, and fire equipment status monitoring data, breaking through the limitations of traditional single data acquisition. Video image data can intuitively display personnel activities within the building, the status of fire protection facilities, and whether there are obstacles blocking passageways. Fire sensor parameters monitor fire hazards at the physical parameter level, such as smoke concentration and temperature changes. Fire equipment status monitoring data provides real-time feedback on the operational status of key fire protection equipment such as fire hydrants and sprinkler systems. This comprehensive fire data collection ensures no omissions in monitoring key aspects of building fire prevention, providing a complete understanding of the building's fire safety situation. Utilizing AI algorithms for deep analysis of dynamic video image data and feature extraction, it can accurately identify personnel, fire protection facilities, and obstacles. For example, it can quickly determine whether fire hydrants are obstructed or whether passageways are cluttered, making it more efficient and precise than manual inspections. Analysis of on-site personnel behavior characteristics can identify abnormal behaviors such as congestion and reverse movement during evacuation, providing timely warnings of potential dangers. Through fire feature extraction… Based on smoke color, shape, and flame characteristics, it can accurately determine whether a fire has occurred, greatly improving the accuracy of fire assessment and reducing false alarms and missed alarms. It integrates video analysis results, fire sensor parameters, and fire equipment status monitoring data, allowing information from different data sources to corroborate and supplement each other, making building fire safety assessments more scientific and comprehensive. Simultaneously, it integrates video target characteristics, on-site personnel behavior characteristics, and fire characteristics to generate fire scene characteristics, providing richer and more accurate data for subsequent decision-making and optimizing the building fire safety assessment process and results. Finally, based on the fire scene characteristics and the final building fire safety inspection results, it quickly generates a fire scene report and sends it to building management personnel and relevant fire departments. This ensures that management personnel can grasp the situation on-site immediately, organize personnel evacuation, and take initial fire-fighting measures. Fire departments can use the report to understand key information such as the fire's intensity and building layout in advance, planning rescue strategies, greatly shortening response time, gaining valuable time for fire fighting, and effectively reducing fire losses.

[0015] Other features and advantages of the invention will be set forth in the description which follows, and will be apparent in part from the description, or may be learned by practicing the invention. The objects and other advantages of the invention may be realized and obtained by means of the structures particularly pointed out in this application.

[0016] The technical solution of the present invention will be further described in detail below with reference to the accompanying drawings and embodiments. Attached Figure Description

[0017] The accompanying drawings are provided to further illustrate the invention and form part of the specification. They are used in conjunction with embodiments of the invention to explain the invention and do not constitute a limitation thereof. In the drawings: Figure 1 This is a structural diagram of a building fire detection data processing system based on multi-source data fusion according to the present invention; Figure 2 This is a structural diagram of the intelligent data analysis module of a building fire detection data processing system based on multi-source data fusion according to the present invention; Figure 3 This is a structural diagram of the multi-source data fusion and decision module of a building fire detection data processing system based on multi-source data fusion according to the present invention; Figure 4 This is a flowchart of a building fire detection data processing method based on multi-source data fusion according to the present invention. Detailed Implementation

[0018] The preferred embodiments of the present invention will be described below with reference to the accompanying drawings. It should be understood that the preferred embodiments described herein are for illustration and explanation only and are not intended to limit the present invention.

[0019] Example 1: This invention provides a building fire detection data processing system based on multi-source data fusion, such as... Figure 1 As shown, it includes: The multi-source data acquisition module is used to collect dynamic video image data, fire sensor parameters, and fire equipment status monitoring data in the building in real time. The data intelligence analysis module is used to extract features from dynamic video image data using AI algorithms to obtain video target features, on-site personnel behavior features, and fire features, determine whether a fire has occurred inside the building, and obtain video analysis results. The video targets include people, fire-fighting equipment, obstacles, and flames; The multi-source data fusion and decision module is used to fuse video analysis results with fire sensor parameters and fire equipment status monitoring data analysis results to obtain the final building fire safety detection results. At the same time, it fuses video target features, on-site personnel behavior features, and fire features to generate fire scene features. The fire report sending module is used to generate a fire scene report based on the characteristics of the fire scene and the final building fire safety inspection results, and send it to building managers and relevant fire departments.

[0020] The beneficial effects of the above technical solution are as follows: This invention simultaneously collects dynamic video image data, fire sensor parameters, and fire equipment status monitoring data, breaking through the limitations of traditional single data collection. Video image data can intuitively display the activities of people in the building, the status of fire-fighting facilities, and whether there are obstacles blocking passages. Fire sensor parameters monitor fire hazards from the physical parameter level, such as smoke concentration and temperature changes. Fire equipment status monitoring data provides real-time feedback on the operating status of key fire-fighting equipment such as fire hydrants and sprinkler systems. Comprehensive fire data collection ensures that no key aspects of building fire prevention are missed, and a comprehensive understanding of the building's fire safety situation is achieved. AI algorithms are used to deeply analyze dynamic video image data and extract video target features, accurately identifying people, fire-fighting facilities, and obstacles. For example, it can quickly determine whether fire hydrants are blocked or whether passages are cluttered with debris, making it more efficient and precise than manual inspection. Analysis of on-site personnel behavior characteristics can identify abnormal behaviors such as congestion and reverse movement during evacuation, providing timely warnings of potential dangers. Disaster feature extraction can accurately determine whether a fire has occurred based on smoke color, shape, and flame characteristics, greatly improving the accuracy of fire assessment and reducing false alarms and missed alarms. It integrates video analysis results, fire sensor parameters, and fire equipment status monitoring data, allowing information from different data sources to corroborate and supplement each other, making building fire safety assessments more scientific and comprehensive. Simultaneously, it generates fire scene features by integrating video target features, on-site personnel behavior characteristics, and fire characteristics, providing richer and more accurate data for subsequent decision-making and optimizing the building fire safety assessment process and results. Finally, based on the fire scene features and the final building fire safety inspection results, it quickly generates a fire scene report and sends it to building management personnel and relevant fire departments. This ensures that management personnel can grasp the situation on-site immediately, organize personnel evacuation, and take initial fire-fighting measures. Fire departments can use the report to understand key information such as the fire's intensity and building layout in advance, planning rescue strategies, greatly shortening response time, gaining valuable time for fire fighting, and effectively reducing fire losses.

[0021] Example 2: Based on Example 1, the data intelligent analysis module, such as Figure 2 As shown, it includes: The video processing unit is used to perform real-time video frame segmentation on dynamic video image data to obtain real-time monitoring images. Based on AI algorithms, it uses rectangles of different colors to mark various video targets in the real-time monitoring images. The fire detection unit is used to determine whether there is a fire marker box in the current video target marking result. If there is, it determines that a fire has occurred in the building, and determines the location of the fire based on the location of the real-time monitoring screen, and generates a fire warning signal to send to the monitoring management terminal. The video feature extraction unit is used to extract features from video targets based on the video target labeling results, thereby obtaining video target features; When the target in the video is identified as a person, the behavior of the person is feature extracted to obtain the behavioral characteristics of the people on site. When it is determined that a fire marker box exists in the current video target marking results, fire features are extracted from the local image within the fire marker box.

[0022] In this embodiment, the rectangles corresponding to different types of video targets are of different colors.

[0023] In this embodiment, the video target features include the video target's position, size, and motion trajectory.

[0024] The beneficial effects of the above technical solution are as follows: This invention performs real-time video frame segmentation on dynamic video image data, enabling the intuitive presentation of the situation inside the building in the form of real-time monitoring footage. Simultaneously, based on AI algorithms, various video targets are marked with rectangles of different colors, allowing monitoring personnel to quickly and clearly identify targets such as personnel, fire-fighting facilities, and obstacles in the image. Subsequently, the presence of fire markers in the video target marking results is used to determine the presence of fire, achieving automatic fire monitoring through AI algorithms and effectively improving fire detection efficiency. Finally, a video feature extraction unit extracts features based on the video target marking results, facilitating the determination of on-site video characteristics. The distribution of video targets (including personnel, fire-fighting facilities, and obstacles) provides a basis for personnel evacuation and firefighting when a fire occurs. When the video target is a person, extracting personnel behavior characteristics can analyze whether there are abnormal behaviors, such as prolonged stays in fire exits or abnormal running directions, and identify potential safety risks in advance. When a fire marker frame is identified, fire features can be extracted from the local image within the fire marker frame, allowing for a more detailed analysis of the fire state, such as the size of the flames and the trend of changes in smoke concentration. The extraction of multi-dimensional features provides rich data support for subsequent fire assessment, fire prediction, and the development of targeted firefighting and rescue strategies, thereby improving the depth and breadth of building fire prevention detection.

[0025] Example 3: Based on Example 2, the data intelligent analysis module, such as Figure 2 As shown, it also includes: The target status detection unit is used to determine the location of each sub-building within the effective area of ​​the building based on the video target marking results, and to locate the fire safety evacuation and fire fighting and rescue facilities within the effective area of ​​the building by combining the fire treatment facility requirements and the building layout data. Based on the location results, determine the location of existing fire safety evacuation and fire fighting and rescue facilities within the effective area of ​​the building, as well as any missing fire safety evacuation and fire fighting and rescue facilities. Based on the missing fire safety evacuation and fire fighting and rescue facilities, generate a fire facility deficiency report and send it to the relevant management personnel. At the same time, based on the image scale of the real-time monitoring screen and the actual image size of the existing fire safety evacuation facilities in the building, the actual reserved location size is calculated, a fire protection facility report is generated and sent to the associated APP for display; In the event of a fire inside a building, a fire safety guidance report is generated and sent to the relevant management personnel.

[0026] In this embodiment, fire safety evacuation and fire fighting and rescue facilities include, but are not limited to, fire separation distances, fire truck access roads, fire truck aerial operation areas, safety exits and evacuation routes, fire hydrants, and fire extinguishers.

[0027] The beneficial effects of the above technical solution are as follows: Based on video target marking results and building layout data, the target state detection unit of this invention can accurately locate fire safety evacuation and fire-fighting rescue facilities within the effective area of ​​a building. Whether it is a large commercial complex or a complex industrial building, it can quickly and accurately determine the location of fire-fighting facilities. It can not only clearly understand the distribution of existing facilities but also promptly identify missing fire-fighting facilities and generate a fire-fighting facility missing report. This helps managers to supplement and improve fire-fighting facilities in a timely manner, ensuring that fire-fighting facilities within the building are fully equipped and rationally laid out, effectively improving the overall fire safety level of the building and avoiding inadequate rescue due to missing or improperly located fire-fighting facilities. Furthermore, by determining the actual reserved dimensions of existing fire-fighting facilities and generating a fire-fighting facility report displayed in the associated APP, managers can intuitively understand the actual dimensions and reserved space of each facility, rationally arrange the maintenance, updating, and addition of facilities, and avoid resource waste and redundant construction. In the event of a fire in a building, the generated fire guidance report can provide key information support to relevant managers, helping firefighters to scientifically plan rescue routes based on real-time fire conditions and improve rescue efficiency.

[0028] Example 4: Based on Example 2, the data intelligent analysis module, such as Figure 2 As shown, it also includes: The duty monitoring unit is used to acquire and analyze video target features and on-site personnel behavior characteristics in the fire control room to determine whether the current number of people on duty in the fire control room is consistent with the prescribed number for the current time period. If there is a discrepancy, an anomaly notification shall be sent to the fire control room manager and the on-duty status of the fire control room shall be monitored. If they match, determine if there is any abnormal behavior by the on-duty personnel. If so, send a behavioral alert voice message to the fire control room. Otherwise, continue to monitor the duty situation in the fire control room.

[0029] The beneficial effects of the above technical solution are as follows: By analyzing the video target characteristics and on-site personnel behavior characteristics in the fire control room, this invention accurately determines whether the current number of on-duty personnel matches the prescribed number, effectively avoiding fire monitoring loopholes caused by personnel absence. For example, during holidays, nighttime, and other periods when insufficient personnel are likely to occur, absences can be promptly detected and abnormal notifications can be sent to the manager, ensuring that the fire control room always has sufficient manpower to respond to sudden fires and avoiding delays in fire warnings and emergency response due to insufficient on-duty personnel. When the number of on-duty personnel meets the requirements, further monitoring can be conducted to check for abnormal behavior (e.g., leaving the post, sleeping, playing on mobile phones, etc.). Immediately, behavioral reminder voice messages can be sent, urging on-duty personnel to correct their behavior in a timely manner, which helps to standardize the work status of on-duty personnel, ensuring that they remain vigilant at all times and minimizing the possibility of fire hazards not being detected and dealt with in a timely manner due to personnel negligence.

[0030] Example 5: Based on Example 1, the multi-source data fusion and decision-making module, such as... Figure 3 As shown, it includes: The decision fusion unit is used to determine whether the detection of each fire sensor is reliable based on the status monitoring data of fire equipment, fuse the analysis data results corresponding to reliable fire sensors with video analysis, determine the range of fire-affected area and the location of the fire point, and generate the final building fire safety detection result based on the range of fire-affected area and the location of the fire point. The on-site video feature fusion unit is used to, when it is determined that there is a fire in the building, take the location of the fire as the target location, and acquire all relevant dynamic video image data of the target location and its corresponding video target features, on-site personnel behavior features, and fire features; Based on the shooting angle information of relevant dynamic video images, the overlapping detection area is determined. Based on the shooting time and the overlapping detection area, the relevant dynamic video image data is stitched together to obtain a panoramic fire image. Based on the panoramic fire image, the video target features, on-site personnel behavior features, and fire features corresponding to all relevant dynamic video image data are fused to obtain panoramic video target features, panoramic on-site personnel behavior features, and panoramic fire features, respectively. The characteristics of the fire scene are obtained by fusing the panoramic video target features, panoramic on-site personnel behavior features, and panoramic fire features.

[0031] The beneficial effects of the above technical solution are as follows: This invention uses a decision fusion unit to assess the reliability of fire sensor detection results based on fire equipment status monitoring data, effectively eliminating false alarms caused by sensor malfunctions, environmental interference, etc. For example, when a smoke sensor alarms, but the pressure of the corresponding area's fire sprinkler system is normal and the fireproof roller shutter door has not received a linkage signal, the system can determine that the smoke sensor detection result is unreliable, avoiding misjudgment of the fire situation. By filtering reliable data and fusing it with video analysis results, the false alarm rate is significantly reduced, and the accuracy of fire detection is significantly improved, ensuring that the detection results truly reflect the building's fire safety status. Moreover, by fusing reliable sensor data with video analysis information, the location of the fire ignition point and the affected area can be determined more accurately. Through cross-verification of multi-source data, combined with the specific location of flames and smoke in the video, and transmission... Information such as temperature and smoke concentration gradients detected by sensors can improve the accuracy of fire location to the meter level, accurately delineate the fire-affected area, and provide a reliable basis for the precise deployment of subsequent fire rescue forces. Then, through the on-site video feature fusion unit, multi-source video data of the target location is integrated when a fire occurs, and a panoramic fire image is generated based on the shooting angle and time information. This breaks through the limitations of a single camera's perspective and fully presents the entire fire scene, including key information such as the distribution of people around the fire source, the status of fire-fighting facilities, and the condition of escape routes. By performing feature fusion on the panoramic image, panoramic video target features, panoramic on-site personnel behavior features, and panoramic fire features are obtained. This enables fire commanders and managers to have a comprehensive and intuitive grasp of the dynamics of the fire scene, providing strong support for the scientific formulation of rescue plans and the organization of personnel evacuation.

[0032] Example 6: Based on Example 5, the on-site video feature fusion unit includes: The first panoramic feature fusion subunit is used to obtain the first image proportion of each relevant dynamic video image data on the fire panoramic image, and to obtain the first fusion weight. Based on the first fusion weight, the video target features and on-site personnel behavior features of all relevant dynamic video images are fused to obtain panoramic video target features and panoramic on-site personnel behavior features. The second panoramic feature fusion subunit is used to obtain the proportion of the second fire center region image of each relevant dynamic video image on the first fire center region image in the fire panoramic image, and to obtain the second fusion weight. Based on the second fusion weight, the fire features of all relevant dynamic video images are fused to obtain panoramic fire features.

[0033] In this embodiment, the first image proportion refers to the proportion of the target area in each relevant dynamic video image on the fire panoramic image; the second image proportion refers to the proportion of the fire center area in each relevant dynamic video image on the fire center area in the fire panoramic image.

[0034] In this embodiment, the fire center area refers to the area where there is open flame or smoke in the monitoring video or panoramic fire image.

[0035] The beneficial effects of the above technical solution are as follows: The present invention, through a first panoramic feature fusion subunit, performs weighted fusion of video target features and on-site personnel behavior features based on image proportion. This refines information on personnel activity and target distribution in different areas. For example, video images in densely populated areas have a higher weight due to their higher proportion, and key information such as evacuation direction and crowding level is more prominent after fusion. This makes the panoramic video target features and panoramic on-site personnel behavior features more closely reflect the actual situation. The second panoramic feature fusion subunit determines the second fusion weight based on the image proportion of the fire center area. Focusing on the image proportion of the fire center area highlights its dominant position in the fusion process, strengthening the extraction and fusion of core fire features such as flame morphology and smoke concentration. This allows for more accurate capture of the essential characteristics of the fire, helping firefighters to more scientifically assess the fire situation and formulate targeted firefighting and rescue strategies.

[0036] Example 7: Based on Example 2, the fire detection unit further includes: The fire analysis and guidance subunit is used to receive and analyze the fire scene characteristics fed back by the multi-source data fusion and decision-making module to determine the optimal fire handling route, including: The safety level determination subunit is used to determine the relative distance between each video target and the fire center area based on the characteristics of the panoramic video targets, and to determine the fire spread speed and fire size according to the fire characteristics. Based on the fire spread rate and fire size, and combined with the relative distance between different video targets and the fire center area, the current safety status of different video targets in the current fire is determined, and based on the current safety status, relatively safe areas and dangerous areas are determined on the panoramic fire image. The on-site evacuation guidance subunit is used to determine the actual movement direction of on-site personnel based on the characteristics of the fire scene, and to determine whether there are personnel moving towards the danger zone at the fire scene; If they exist, the personnel on site will be divided into multiple evacuation directions based on the location of the safety escape exits, and the name of the landmark item in each direction and its relative location data with the corresponding escape route will be determined. Based on the names of iconic objects and their relative locations to the corresponding escape routes, combined with the road division data within the building, voice broadcast information guiding the evacuation routes in the corresponding evacuation directions is generated and broadcast.

[0037] The beneficial effects of the above technical solution are as follows: This invention, through a safety level determination subunit, determines the relative distance between each video target and the fire center area based on the characteristics of the panoramic video target. Simultaneously, by combining fire characteristics (fire spread speed, fire size), it can accurately assess the safety status of different video targets (such as personnel, important equipment, etc.) in the current fire, and clearly delineate relatively safe and dangerous areas on the panoramic fire image. This provides a reliable basis for subsequent evacuation decisions, greatly improving the ability of management and fire personnel to grasp the safety situation at the fire scene. Subsequently, through the on-site evacuation guidance subunit, based on the characteristics of the fire scene… The system determines the actual movement direction of on-site personnel and assesses whether anyone is moving towards a danger zone. Based on the location of safety escape exits, personnel are divided into multiple evacuation directions. It also identifies landmarks and their relative positions to escape routes. By fully considering the actual conditions of the fire scene and the distribution of personnel, the system plans on-site evacuation trajectories. This prevents personnel from blindly moving to danger zones during evacuation, effectively improving evacuation efficiency and safety. Furthermore, it generates and broadcasts voice guidance information for corresponding evacuation directions, enabling more timely and effective communication of evacuation instructions and reducing confusion and disarray during the evacuation process.

[0038] Example 8: Based on Example 7, the fire analysis guidance subunit further includes: The intelligent rescue guidance subunit is used to identify on-site personnel moving towards the danger zone as rescue targets based on the characteristics of the fire scene, and to mark, track, and count them to obtain the real-time dynamic trajectory and number of rescue targets. The marking and tracking are removed after the rescue targets leave the danger zone. Based on the location of the boundary between the dangerous area and the relatively safe area, determine the available fire-fighting equipment nearby, and control the sound positioning alarm on the available fire-fighting equipment to emit a warning sound; At the same time, based on the characteristics of the fire scene, the weak points of the fire in the central area of ​​the fire are determined. Based on the weak points and the real-time dynamic trajectory of the target to be rescued, multiple recommended priority rescue routes are obtained. After marking the locations of all targets to be rescued on each priority rescue route, the information is sent to firefighters or building managers.

[0039] In this embodiment, the recommended priority rescue route for the target to be rescued, which is actually collected by the video surveillance equipment, has a higher priority than the recommended priority rescue route for the target to be rescued generated based on the alternative trajectory. This provides a foundation for achieving the maximum number of accurate rescues in the shortest time.

[0040] The beneficial effects of the above technical solution are as follows: This invention treats on-site personnel moving towards dangerous areas as targets to be rescued, and marks, tracks, and counts them. It can obtain the dynamic trajectory and number of these targets in real time, allowing firefighters and building managers to clearly understand their location and movement trajectory, accurately determine their location, provide precise information support for rescue operations, improve the targeting and effectiveness of rescue efforts, effectively avoid omissions and misjudgments, and, based on the boundary between dangerous and relatively safe areas, identify nearby available fire-fighting equipment and control its sound-based location alarm to emit a warning sound, guiding firefighters and trapped personnel to quickly locate the equipment, thereby improving... The equipment's high efficiency provides strong support for firefighting and rescue operations. Simultaneously, based on the characteristics of the fire scene, it identifies the weak points in the fire's central area and, combined with the real-time dynamic trajectories of the targets to be rescued, obtains multiple recommended priority rescue routes. The locations of the targets to be rescued are marked on each priority rescue route and sent to firefighters and building managers, enabling them to clearly understand the distribution of the targets and the conditions of the rescue routes. This provides firefighters and building managers with relatively safe and efficient rescue paths, effectively improving the safety and success rate of rescue operations, greatly increasing the chances of rescuing trapped personnel, and reducing the risk of casualties caused by fires.

[0041] Example 9: Based on Example 8, the fire analysis guidance subunit further includes: The target trajectory analysis subunit is used to obtain the actual running trajectory and moving speed of the target on the panoramic image of the fire scene when the video surveillance equipment at the fire scene is damaged and the tracking of the target fails. Based on the actual running trajectory and movement speed, the running trajectory of the target to be rescued is predicted, the predicted trajectory is obtained, and the predicted trajectory is sent to the intelligent rescue guidance subunit as a substitute trajectory for the real-time movement trajectory of the target to be rescued.

[0042] The beneficial effects of the above technical solution are as follows: When the video monitoring equipment at the fire scene is damaged (for example, the smoke or explosion at the fire scene causes damage to some cameras) and the tracking of the target to be rescued fails, the target trajectory analysis subunit can obtain the actual running trajectory and moving speed of the target to be rescued on the panoramic image of the fire, predict the running trajectory of the target to be rescued and generate a predicted trajectory, and send the predicted trajectory as a substitute trajectory for the real-time movement trajectory of the target to be rescued to the intelligent rescue guidance subunit, thereby improving the foundation for rescue personnel (including managers and firefighters) to plan rescue routes.

[0043] Example 10: This invention provides a method for processing building fire detection data based on multi-source data fusion, such as... Figure 4 As shown, it includes: Step 1: Real-time acquisition of dynamic video image data, fire sensor parameters, and fire equipment status monitoring data within the building; Step 2: Use AI algorithms to extract features from dynamic video image data to obtain video target features, on-site personnel behavior features, and fire features, determine whether a fire has occurred inside the building, and obtain video analysis results; The video targets include people, fire-fighting equipment, obstacles, and flames; Step 3: Integrate the video analysis results with the fire sensor parameters and the fire equipment status monitoring data analysis results to obtain the final building fire safety test results. At the same time, integrate the video target features, on-site personnel behavior features and fire features to generate fire scene features. Step 4: Based on the characteristics of the fire scene and the final building fire safety inspection results, generate a fire scene report and send it to building management personnel and relevant fire departments.

[0044] The beneficial effects of the above technical solution are as follows: This invention simultaneously collects dynamic video image data, fire sensor parameters, and fire equipment status monitoring data, breaking through the limitations of traditional single data collection. Video image data can intuitively display the activities of people in the building, the status of fire-fighting facilities, and whether there are obstacles blocking passages. Fire sensor parameters monitor fire hazards from the physical parameter level, such as smoke concentration and temperature changes. Fire equipment status monitoring data provides real-time feedback on the operating status of key fire-fighting equipment such as fire hydrants and sprinkler systems. Comprehensive fire data collection ensures that no key aspects of building fire prevention are missed, and a comprehensive understanding of the building's fire safety situation is achieved. AI algorithms are used to deeply analyze dynamic video image data and extract video target features, accurately identifying people, fire-fighting facilities, and obstacles. For example, it can quickly determine whether fire hydrants are blocked or whether passages are cluttered with debris, making it more efficient and precise than manual inspection. Analysis of on-site personnel behavior characteristics can identify abnormal behaviors such as congestion and reverse movement during evacuation, providing timely warnings of potential dangers. Disaster feature extraction can accurately determine whether a fire has occurred based on smoke color, shape, and flame characteristics, greatly improving the accuracy of fire assessment and reducing false alarms and missed alarms. It integrates video analysis results, fire sensor parameters, and fire equipment status monitoring data, allowing information from different data sources to corroborate and supplement each other, making building fire safety assessments more scientific and comprehensive. Simultaneously, it generates fire scene features by integrating video target features, on-site personnel behavior characteristics, and fire characteristics, providing richer and more accurate data for subsequent decision-making and optimizing the building fire safety assessment process and results. Finally, based on the fire scene features and the final building fire safety inspection results, it quickly generates a fire scene report and sends it to building management personnel and relevant fire departments. This ensures that management personnel can grasp the situation on-site immediately, organize personnel evacuation, and take initial fire-fighting measures. Fire departments can use the report to understand key information such as the fire's intensity and building layout in advance, planning rescue strategies, greatly shortening response time, gaining valuable time for fire fighting, and effectively reducing fire losses.

[0045] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A building fire prevention detection data processing system based on multi-source data fusion, characterized in that, The application relates to a building fire safety detection system and method. The system comprises: a multi-source data acquisition module for acquiring dynamic video image data, fire sensing parameters and fire equipment state monitoring data in a building in real time; a data intelligent analysis module for extracting features from the dynamic video image data by using an AI algorithm to obtain video target features, on-site personnel behavior features and fire features, and judging whether a fire occurs in the building to obtain a video analysis result; wherein the video target comprises personnel, fire-fighting facilities, obstacles and flames; a multi-source data fusion and decision module for fusing the video analysis result with the fire sensing parameters and the fire equipment state monitoring data analysis result to obtain a final building fire safety detection result, and fusing the video target features, the on-site personnel behavior features and the fire features to generate a fire scene feature; 2. The building fire prevention detection data processing system based on multi-source data fusion according to claim 1, characterized in that, a fire report sending module for generating a fire scene report based on the fire scene feature and the final building fire safety detection result and sending the report to building managers and relevant fire departments. The data intelligent analysis module comprises: a video processing unit for performing real-time video frame segmentation on the dynamic video image data to obtain a real-time monitoring picture, and marking multiple video targets in the real-time monitoring picture by using different color rectangular frames based on an AI algorithm; a fire judgment unit for judging whether a fire marking frame exists in a current video target marking result, and if the fire marking frame exists, judging that a fire occurs in the building, determining a fire occurrence position according to a real-time monitoring picture acquisition position, and generating a fire warning signal and sending the signal to a monitoring management end; a video feature extraction unit for extracting features from the video target based on the video target marking result to obtain video target features; when the video target is determined as a person, behavior features of the person are extracted to obtain on-site personnel behavior features; 3. The building fire prevention detection data processing system based on multi-source data fusion according to claim 2, characterized in that, when it is determined that the fire marking frame exists in the current video target marking result, fire features of a local picture in the fire marking frame are extracted. The data intelligent analysis module further comprises: a target state detection unit for determining positions of each sub-building in a building effective area based on the video target marking result, positioning fire safety evacuation and fire rescue facilities in the building effective area in combination with fire handling facility requirements and building layout data, determining existing fire safety evacuation and fire rescue facility positions and missing fire safety evacuation and fire rescue facilities in the building effective area according to the positioning result, and generating a fire facility missing report according to the missing fire safety evacuation and fire rescue facilities and sending the report to relevant managers; at the same time, actual reserved position sizes of the existing fire safety evacuation facilities in the building are calculated based on an image scale of the real-time monitoring picture and actual image sizes of the existing fire safety evacuation facilities in the building, a fire facility report is generated and sent to an associated APP for display; 4. The building fire prevention detection data processing system based on multi-source data fusion according to claim 2, characterized in that, when a fire occurs in the building, a fire guidance report is generated and sent to relevant managers. The data intelligent analysis module further comprises: a duty monitoring unit for acquiring video target features and on-site personnel behavior features of a fire control room for analysis, judging whether a current number of on-duty persons in the fire control room is consistent with a specified number corresponding to a current time period, If not, send an abnormality notification to the fire control room manager and keep monitoring the on-duty situation of the fire control room; If yes, determine whether the on-duty personnel have abnormal behavior, and if yes, send a behavior reminder voice to the fire control room; Otherwise, continue to monitor the on-duty situation of the fire control room.

5. The building fire prevention detection data processing system based on multi-source data fusion according to claim 1, characterized in that, The multi-source data fusion and decision module comprises: A decision fusion unit is configured to determine whether the detection of each fire sensor is reliable based on the fire equipment state monitoring data, fuse the analysis data results of the reliable fire sensors and the video analysis, determine the fire influence area range and the fire point position, and generate a final building fire safety detection result based on the fire influence area range and the fire point position; A field video feature fusion unit is configured to, when it is determined that there is a fire in the building, take the fire occurrence position as a target position, and acquire all related dynamic video image data of the target position and corresponding video target features, field personnel behavior features, and fire features; According to shooting angle information of the related dynamic video images, determine a detection overlap area based on the shooting angle information, splice the related dynamic video image data based on shooting time and the detection overlap area, and acquire a fire panoramic image; Fuse the video target features, field personnel behavior features, and fire features corresponding to all related dynamic video image data based on the fire panoramic image, and respectively acquire panoramic video target features, panoramic field personnel behavior features, and panoramic fire features; And fuse the panoramic video target features, panoramic field personnel behavior features, and panoramic fire features to acquire fire scene features.

6. The building fire prevention detection data processing system based on multi-source data fusion according to claim 5, characterized in that, The field video feature fusion unit comprises: A first panoramic feature fusion subunit is configured to acquire a first image proportion of each related dynamic video image data on the fire panoramic image, and obtain a first fusion weight; Fuse the video target features and field personnel behavior features of all related dynamic video images based on the first fusion weight, and obtain panoramic video target features and panoramic field personnel behavior features; A second panoramic feature fusion subunit is configured to acquire a second image proportion of a second fire center area image of each related dynamic video image on the first fire center area image on the fire panoramic image, and obtain a second fusion weight; Fuse the fire features of all related dynamic video images according to the second fusion weight, and obtain panoramic fire features.

7. The building fire prevention detection data processing system based on multi-source data fusion according to claim 2, characterized in that, The fire condition determination unit further comprises: A fire condition analysis guidance subunit is configured to receive the fire scene features fed back by the multi-source data fusion and decision module and perform analysis, and determine a best fire handling route, comprising: A safety level determination subunit is configured to determine relative distances between each video target and the fire center area based on the panoramic video target features, determine a fire spread speed and a fire size according to the fire features, determine current safety conditions of different video targets in the current fire according to the fire spread speed and the fire size in combination with the relative distances between the different video targets and the fire center area, and determine relatively safe areas and dangerous areas on the fire panoramic image based on the current safety conditions. ​ The on-site evacuation guidance subunit is configured to determine the actual moving direction of the on-site personnel according to the fire scene characteristics, and determine whether there are personnel moving towards the dangerous area in the fire scene; If so, the on-site personnel are divided into multiple evacuation directions based on the location of the safe escape exit, and the name of the landmark object and the relative position data of the landmark object and the corresponding escape passage in each direction are determined; Based on the name of the landmark object and the relative position data of the landmark object and the corresponding escape passage, combined with the road division data in the building, the personnel evacuation route guidance voice broadcast information corresponding to the evacuation direction is generated and broadcasted.

8. The building fire prevention detection data processing system based on multi-source data fusion according to claim 7, characterized in that, The fire analysis guidance subunit further comprises: The intelligent rescue guidance subunit is configured to determine the on-site personnel moving towards the dangerous area as the to-be-rescued target based on the fire scene characteristics, and mark and track the to-be-rescued target to obtain the real-time dynamic trajectory of the to-be-rescued target and the real-time number of to-be-rescued targets, and eliminate the marking and tracking after the to-be-rescued target leaves the dangerous area; Based on the location of the critical line between the dangerous area and the relatively safe area, the available fire-fighting equipment is determined, and the sound positioning alarm on the available fire-fighting equipment is controlled to emit a reminder sound; At the same time, the weak position of the fire in the fire center area is determined according to the fire scene characteristics, and based on the weak position, combined with the real-time dynamic trajectory of the to-be-rescued target, a plurality of recommended priority rescue routes are obtained; After marking the positions of all the to-be-rescued targets on each priority rescue route, the positions are sent to the fire-fighting personnel or the building management personnel.

9. The building fire prevention detection data processing system based on multi-source data fusion according to claim 8, characterized in that, The fire analysis guidance subunit further comprises: The target trajectory analysis subunit is configured to, when the video monitoring device of the fire scene is damaged, causing the tracking of the to-be-rescued target to fail, obtain the actual running trajectory and moving speed of the to-be-rescued target on the fire panoramic image; Based on the actual running trajectory and moving speed, the running trajectory of the to-be-rescued target is predicted to obtain a predicted trajectory, and the predicted trajectory is sent to the intelligent rescue guidance subunit as a substitute trajectory of the real-time action trajectory of the to-be-rescued target.

10. A building fire prevention detection data processing method based on multi-source data fusion, characterized in that, The method comprises: Step 1: real-time collection of dynamic video image data, fire sensing parameters and fire-fighting equipment state monitoring data in the building; Step 2: feature extraction of the dynamic video image data by using an AI algorithm to obtain video target features, on-site personnel behavior features and fire characteristics, to determine whether a fire occurs in the building, and to obtain a video analysis result; The video target comprises personnel, fire-fighting facilities, obstacles and fire; Step 3: fusion of the video analysis result, the fire sensing parameters and the fire-fighting equipment state monitoring data analysis result to obtain a final building fire safety detection result, and fusion of the video target features, the on-site personnel behavior features and the fire characteristics to generate fire scene characteristics; Step 4: generation of a fire scene report based on the fire scene characteristics and the final building fire safety detection result, and sending of the report to the building management personnel and the relevant fire department.

Images