Unmanned aerial vehicle urban high-rise building fire detection method based on multispectral fusion algorithm

Abstract

The invention relates to the field of fire detection, in particular to an unmanned aerial vehicle urban high-rise building fire detection method based on a multispectral fusion algorithm, which comprises the following steps: converting multi-source data into layered detection parameters, generating a task instruction and issuing the task instruction to an unmanned aerial vehicle; collecting multi-modal data according to a task instruction, preprocessing the multi-modal data, and then returning the data in real time; when any modal data is detected to be abnormal, a reconnaissance instruction is issued, and the unmanned aerial vehicle collects reconnaissance data and returns the reconnaissance data; processing the reconnaissance data through an improved multispectral fusion algorithm, completing fire behavior judgment in combination with a preset high-rise building exclusive dynamic combustible library, and generating a judgment result; and generating a standardized fire behavior report based on a judgment result, transmitting the standardized fire behavior report to a fire-fighting command system through a fire-fighting linkage interface, and updating the dynamic combustible library after the task is completed. Layered detection, multi-modal fusion and the dynamic combustible material library are taken as the core, multiple systems are cooperated, the fire detection comprehensiveness and accuracy of the high-rise building are improved, and efficient linkage rescue is achieved.

Description

Technical Field

[0001] This invention relates to the field of fire detection, and more particularly to a method for detecting fires in urban high-rise buildings using a multispectral fusion algorithm via unmanned aerial vehicles (UAVs). Background Technology

[0002] With the deep integration of drone technology and spectral detection technology, drone fire detection has become an important development direction in the fire protection field, gradually making up for the limitations of traditional fixed detection equipment. Currently, drone fire detection solutions on the market are mainly divided into two categories: one is a detection method based on a single spectral sensor, which uses an infrared sensor or visible light camera to achieve preliminary identification and location of the fire area. Leveraging the maneuverability of drones, they can quickly reach areas that are difficult for traditional equipment to cover. The other category is a detection solution that incorporates multispectral fusion technology, integrating multimodal data such as infrared, ultraviolet, and thermal imaging, combined with a pre-set combustible combustion characteristic database, to determine whether a fire has occurred and the type of combustible material. Compared with single-spectral detection, this method offers a certain improvement in accuracy. Meanwhile, some ground-based fixed multispectral detection systems have also been applied to building fire monitoring. By collecting environmental data through fixed multispectral sensors, they provide a reference for fire early warning. These technologies have demonstrated certain application value in ordinary buildings and open areas.

[0003] However, existing technologies still face many unresolved issues in detecting fires in urban high-rise buildings. In terms of detection coverage, conventional drone detection lacks specific design for the three-dimensional structure of high-rise buildings, and lacks targeted detection plans for fire-prone areas at different height levels, such as low-rise windows and balconies, mid-rise curtain wall joints, and high-rise rooftop equipment areas, making it difficult to achieve comprehensive, blind-spot-free coverage. Ground-based fixed multispectral detection systems are limited by deployment location, unable to overcome spatial constraints, and cannot obtain accurate detection data for high-altitude areas within high-rise buildings. Regarding accuracy, existing multispectral fusion algorithms do not dynamically adjust the fusion strategy based on environmental differences at different height levels of high-rise buildings, resulting in insufficient anti-interference capabilities. Furthermore, combustible material databases are mostly statically pre-set, failing to incorporate the combustion characteristics of new structural and decorative materials in high-rise buildings in a timely manner, leading to significant errors in the judgment of fires involving special materials. From the perspective of coordinated response, the existing solutions lack a standardized docking mechanism with the urban fire command system. The process of converting fire detection results into rescue instructions is cumbersome, and there is no priority sorting function designed for multiple fire scenarios. This makes it difficult to support efficient rescue deployment and fails to meet the high requirements of comprehensiveness, accuracy, rapid response, and efficient coordination for fire detection in urban high-rise buildings. Summary of the Invention

[0004] To address the technical deficiencies in the background technology, this invention proposes a UAV-based method for detecting fires in high-rise buildings in urban areas using a multispectral fusion algorithm. This method solves the aforementioned technical problems and meets practical needs. The specific technical solution is as follows: A method for detecting fires in high-rise buildings in cities using unmanned aerial vehicles (UAVs) based on multispectral fusion algorithms includes the following steps: The ground control center receives the address, height, structural drawings, and personnel distribution information of high-rise buildings, converts them into layered detection parameters, generates mission instructions, and sends them to the UAVs equipped with multimodal spectral detection units. The UAV collects multimodal data in layers according to mission instructions, and transmits the pre-processed data back to the ground control center in real time. When the ground control center detects any abnormality in any modal data, it issues a reconnaissance command, and the UAV switches to reconnaissance mode to collect reconnaissance data at close range and transmit it back. The ground control center processes reconnaissance data using an improved multispectral fusion algorithm, and combines it with a pre-set dynamic combustible material warehouse specific to high-rise buildings to complete the fire situation assessment and generate the assessment result; A standardized fire report is generated based on the judgment results and transmitted to the fire command system through the fire linkage interface. The dynamic combustible material inventory is updated after the task is completed.

[0005] Furthermore, the specific steps for generating the task instructions are as follows: The ground control center analyzes the received information about high-rise buildings, divides them into different height levels according to their height, and clarifies the coverage area of ​​each level. Based on the building structure drawings, identify fire-prone areas on each floor, and plan detection path patterns according to the structural characteristics of different areas; Based on the characteristics of the building's surrounding environment and historical meteorological data, appropriate sensor acquisition frequencies and drone cruising speed parameters are set for each level. By integrating the results of the hierarchical classification, the detection path, the acquisition frequency, and the cruise speed, standardized hierarchical detection mission instructions are generated.

[0006] Furthermore, the specific steps of the preprocessing are as follows: The edge computing unit onboard the drone receives multimodal data and environmental parameters collected by the multimodal spectral detection unit and the environmental perception unit; Based on multimodal data and environmental parameters, corresponding preprocessing algorithms are used for noise filtering, signal enhancement, and data smoothing for different types of data. The preprocessed multimodal data is associated with environmental parameters by timestamp to form structured data; Structured data is compressed and transmitted back to the ground control center in real time via a communication module to ensure the real-time performance and integrity of data transmission.

[0007] Furthermore, the specific steps for transmitting data to the fire command system via the fire alarm linkage interface are as follows: The ground control center integrates fire situation assessment results, type of burning material, fire level, and priority ranking information of multiple fire points, and generates a standardized fire report according to the preset report format. Through a standardized linkage interface compatible with general fire communication protocols, fire reports are transmitted to the city's fire command system, triggering dispatch and related emergency linkage commands; The drones maintained a patrol around the fire area, continuously collecting data on the dynamic changes in the fire situation and transmitting it back to the ground control center in real time. Based on the dynamic data transmitted back, the ground control center regularly updates the fire report, synchronizes it with the fire command system, and generates targeted prevention and control recommendations according to the type of combustible material.

[0008] Furthermore, the specific process for collecting the reconnaissance data is as follows: The ground control center and the edge computing unit carried by the UAV establish a collaborative analysis mechanism to jointly determine anomalies in the preprocessed multimodal data; When any modal data reaches the preset anomaly judgment standard, the ground control center immediately issues a reconnaissance trigger command to the UAV; After receiving instructions, the drone switches to reconnaissance mode and uses the positioning module to accurately locate the spatial coordinates of the abnormal area. Adjust the flight path, control the distance from the abnormal area, reduce the flight speed, increase the acquisition frequency of each sensor, and enable the visible light video recording function. It continuously collects multimodal data and environmental parameters, associates them according to preset rules to form a complete reconnaissance data packet, and transmits it back to the ground control center in real time through a high-speed communication network.

[0009] Furthermore, the specific steps for processing reconnaissance data using the improved multispectral fusion algorithm are as follows: After receiving the reconnaissance data packets transmitted back by the UAV, the ground control center extracts the environmental parameters, including temperature, humidity, wind speed, and air pressure data. Based on the height and hierarchical characteristics of high-rise buildings, the weighting factors of each spectral mode, including infrared, ultraviolet, thermal imaging, visible light, and smoke, are determined, and the confidence level of each mode is dynamically adjusted according to temperature and humidity parameters. A weighted fusion method is used to integrate and calculate the multimodal data in the reconnaissance data packet, eliminating the limitations of single-modal data and generating comprehensive fire feature values; The comprehensive fire characteristic values ​​were compared one by one with the combustion characteristic data of various materials in the dedicated dynamic combustible material warehouse of high-rise buildings; The fire situation is determined based on the comparison results. If the fire situation is confirmed, the type of combustible material is further identified, and the fire level is determined in combination with relevant combustion characteristic indicators.

[0010] Furthermore, the construction and updating of the dedicated dynamic combustible material storage facility for high-rise buildings shall be carried out according to the following steps: Collect common materials for high-rise buildings, covering building structural materials, decorative materials, pipe materials, and electrical equipment-related materials; determine the spectral characteristics and combustion-related parameters of various materials under different environmental conditions; and construct a basic combustible material library. When the spectral data collected by the UAV during the exploration process cannot be matched with the data in the basic combustible material database, the edge computing unit extracts the spectral feature value of the unknown material and associates it with the environmental parameters and visible light image features at the time of collection. The extracted data related to unknown materials is uploaded to the ground control center via a communication network, and then synchronized to the city's fire protection big data platform. The platform assigns professional reviewers to review the data, make a preliminary judgment on the material properties, and after the review is passed, start a simulated combustion experiment to determine the complete combustion characteristic parameters of the material under different environmental conditions. After the experimental data is verified to be correct, it is submitted to the dynamic combustible material storage management system, and the system automatically updates the data in the storage.

[0011] Furthermore, when multiple fire points are identified, it is necessary to calculate the priority ranking of each fire point. The specific steps are as follows: The ground control center collects the positioning data and multiple frames of visible light image data transmitted back by the UAV; A 3D modeling algorithm is used to process the positioning data and multi-frame visible light image data to generate a complete 3D model of the high-rise building. Accurately mark the spatial location, burning range, and spread information of each fire point in the 3D model of the high-rise building; Establish an assessment model that includes three dimensions: personnel density, combustible material hazard, and spread risk, and clarify the evaluation criteria for each dimension; The relevant data of each fire point are input into the evaluation model to calculate the comprehensive score, and different priority levels are divided according to the score results.

[0012] Furthermore, the UAV is equipped with a multimodal spectral detection unit and an environmental perception unit, and features automatic calibration and low-power coordination, as detailed below: Before each reconnaissance mission, the drone initiates a self-check and calibration procedure at the ground take-off and landing point, calling up preset standard reference objects; The multimodal spectral detection unit and the environmental sensing unit respectively collect relevant data from the standard reference object and compare them with the preset standard data; Calculate the deviation between the data collected by various sensors and the standard data, and automatically adjust the sensor's detection parameters when the deviation exceeds the allowable range; After calibration, the drone takes off to perform a detection mission. If no abnormal data is detected, it switches to a low-power operation mode and adjusts the cruise speed and sensor acquisition frequency. When abnormal data is detected, the system automatically exits low-power mode and restores normal operating parameters to ensure the accuracy of abnormal area detection.

[0013] Furthermore, the review and approval process for updating the dynamic combustible material inventory shall be carried out in accordance with the following steps: The urban fire protection big data platform receives data related to unknown materials synchronously from the ground control center, including spectral characteristic values, acquisition environment parameters, and visible light image characteristics. The platform distributes data to qualified auditors according to preset rules, and the auditors independently review and judge the attributes of the materials. Based on the review results of multiple auditors, it was determined whether to initiate a simulated combustion experiment to measure the combustion characteristic parameters of the material. After the experiment is completed, professionals will verify the experimental data and generate a complete report on the material combustion characteristics. The verified combustion characteristic parameters are entered into the dynamic combustible material database, and relevant information such as update time, reviewer, and experimenter is recorded to form an update log.

[0014] Compared with existing technologies, the UAV-based urban high-rise building fire detection method using a multispectral fusion algorithm provided by this invention has the following advantages: This invention significantly improves the overall performance of fire detection in urban high-rise buildings by constructing a collaborative detection system of "ground control center - UAV - fire command system" with layered detection, multimodal data fusion and dynamic combustible material inventory updates as its core. By employing a layered detection mission planning and close-range anomaly reconnaissance mechanism, comprehensive coverage and precise focusing of fire-prone areas at different height levels of high-rise buildings are achieved, effectively overcoming the coverage limitations of traditional fixed detection methods. An improved multispectral fusion algorithm dynamically adjusts weighting factors and modal confidence levels based on height-level characteristics and environmental parameters, coupled with a closed-loop update mechanism for a dynamic combustible material database specific to high-rise buildings, significantly improving the accuracy and adaptability of fire assessment and reducing the risk of misjudgment and missed detection. Through a standardized fire linkage interface and dynamic fire update mechanism, seamless integration of detection data and fire rescue commands is achieved. The multi-fire point priority ranking function provides a scientific basis for rescue decisions, significantly improving rescue efficiency. Automatic calibration and low-power collaborative design of the UAV-mounted equipment ensure the reliability of detection data and system endurance. The overall solution balances comprehensive detection, accurate assessment, efficient linkage, and convenient operation and maintenance, fully meeting the practical needs of fire detection in urban high-rise buildings. Attached Figure Description

[0015] Figure 1 This is a flowchart illustrating the UAV-based urban high-rise building fire detection method using a multispectral fusion algorithm, as described in this invention.

[0016] Figure 2 This is a schematic diagram of the process for calculating the priority ranking of each fire point in this invention. Detailed Implementation

[0017] In the description of this invention, it should be understood that the terms "upper," "lower," "front," "rear," "left," "right," "vertical," "horizontal," "middle," and "inner," etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings. They are used only for the convenience of describing the invention and for simplifying the description, and do not indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation. Therefore, they should not be construed as limitations on the invention. Furthermore, the terms "first," "second," etc., are used for descriptive purposes only and should not be construed as indicating or implying relative importance or implicitly specifying the number of indicated technical features. Thus, features defined with "first," "second," etc., may explicitly or implicitly include one or more of that feature. In the description of this invention, it should be noted that unless otherwise explicitly specified and limited, the terms "installed," "connected," and "joined" should be interpreted broadly. For example, they can refer to a fixed connection, a detachable connection, or an integral connection; they can refer to a direct connection or an indirect connection through an intermediate medium; they can refer to the internal communication of two components. Those skilled in the art can understand the specific meaning of the above terms in this invention through specific circumstances.

[0018] The embodiments of the present invention will be described below with reference to the accompanying drawings and related examples. The embodiments of the present invention are not limited to the following examples, and the present invention relates to the relevant necessary components in this technical field, which should be regarded as well-known technology in this technical field and can be known and mastered by those skilled in this technical field.

[0019] See Figure 1 This invention provides a method for detecting fires in urban high-rise buildings using a multispectral fusion algorithm via unmanned aerial vehicles (UAVs), comprising the following steps: Step S100: The ground control center receives the address, height, structural drawings and personnel distribution information of the high-rise building, converts them into layered detection parameters, generates mission instructions and sends them to the UAV equipped with the multimodal spectral detection unit; The ground control center is a central control unit responsible for receiving information from high-rise buildings, generating detection mission instructions, processing data transmitted back by UAVs, and executing fire situation assessment and linkage transmission. It can be used to schedule and manage UAV detection missions, perform multimodal data fusion analysis, make fire situation assessment decisions, and facilitate standardized interaction with external systems. Layered detection parameters are a set of mission planning parameters derived from the address, height, structural drawings, and personnel distribution information of high-rise buildings. These parameters can guide UAVs to perform systematic detection missions at different altitude levels. The multimodal spectral detection unit is an integrated acquisition unit that combines infrared sensors, ultraviolet sensors, thermal imagers, visible light cameras, and smoke sensors. It can simultaneously acquire spectral data, thermal field data, image data, and smoke concentration data of the target area, providing multi-dimensional data support for fire detection.

[0020] Step S200: The UAV collects multimodal data in layers according to the mission instructions, and transmits the pre-processed data back to the ground control center in real time. The UAV's onboard control unit plans layered flight paths based on the layered parameters in the mission instructions, and sequentially conducts data acquisition according to preset altitude levels. The multimodal spectral detection unit simultaneously activates infrared sensors, ultraviolet sensors, thermal imagers, high-definition visible light cameras, and smoke sensors to collect infrared spectral data, ultraviolet spectral data, thermal field distribution data, scene image data, and smoke concentration data for the corresponding altitude levels. Simultaneously, the environmental perception unit activates temperature and humidity sensors, wind speed sensors, and barometric pressure sensors to collect environmental parameter data for the current altitude level in real time. After acquisition, the onboard edge computing unit immediately preprocesses the raw data: an adaptive filtering algorithm is used to filter random noise in the spectral data, image enhancement technology is used to improve the clarity of visible light images, and a data smoothing algorithm is used to eliminate interference from instantaneous fluctuations in smoke concentration and environmental parameters. The preprocessed multimodal data and environmental parameters are linked and bound by timestamps to form structured data blocks, which are then transmitted back to the ground control center in real time via a high-speed communication link to ensure the continuity and integrity of data transmission.

[0021] Step S300: When the ground control center detects any abnormality in the modal data, it issues a reconnaissance command, and the UAV switches to reconnaissance mode to collect reconnaissance data at close range and transmits it back. Reconnaissance mode is a high-density sampling operation triggered by an anomaly in a specific modality of data. It can be used to improve the data resolution of suspicious areas to verify potential fires. Reconnaissance data consists of multimodal spectral data collected densely at close range in reconnaissance mode, providing higher signal-to-noise ratio and spatial accuracy for precise determination.

[0022] Step S400: The ground control center processes the reconnaissance data using an improved multispectral fusion algorithm, and combines it with the preset dynamic combustible material warehouse for high-rise buildings to complete the fire situation determination and generate the determination result. The improved multispectral fusion algorithm is an optimized and upgraded model based on traditional multispectral fusion technology. It dynamically adjusts the weights and confidence levels of each modality by combining environmental parameters and hierarchical characteristics, generating comprehensive feature values ​​through weighted fusion calculations, thus achieving deep integration and efficient utilization of multi-source data. The dynamic combustible material database specifically for high-rise buildings is a dynamic database built for common materials in high-rise buildings. It stores parameters such as combustion spectral characteristics, thermal field characteristics, and smoke release characteristics of various building structural materials, decorative materials, pipe materials, and electrical equipment-related materials under different environmental conditions such as temperature, humidity, and air pressure. It supports continuous updates and improvements based on detection data feedback. Fire determination is the process of confirming the existence and type of a real fire based on the matching results of the fused reconnaissance data and the dynamic combustible material database. It can be used to output conclusions about whether a fire has occurred and the type of combustible materials.

[0023] Step S500: Generate a standardized fire report based on the judgment result, transmit it to the fire command system through the fire linkage interface, and update the dynamic combustible material inventory after the task is completed.

[0024] Standardized fire reports are fire information documents organized according to a unified format, including elements such as time, location, severity level, and recommended response measures. They facilitate automatic parsing and processing by the fire command system. The fire linkage interface is an interface module designed in accordance with national fire communication standards and industry specifications. It enables data interaction and command transmission between the fire detection system and external systems such as the urban fire command system and emergency rescue platforms, ensuring rapid sharing of fire information and efficient issuance of rescue commands. The fire command system serves as an information management system for urban fire departments to receive alarms, dispatch resources, and formulate rescue plans. It can be used to receive fire reports and initiate emergency response procedures. The dynamic combustible material database update mechanism is an operational process that writes newly detected combustion characteristics back into the high-rise building's dedicated dynamic combustible material database after a task is completed. This allows for continuous evolution and improved adaptability of the knowledge base.

[0025] Taking nighttime fire monitoring of high-rise office buildings as an example, the UAV urban high-rise building fire detection method based on the multispectral fusion algorithm of this invention can detect a localized smoldering fire caused by a short circuit in a 50-story office building at night. The ground control center divides the building into three detection levels based on its height: low (1-15 floors), middle (16-35 floors), and high (36-50 floors). The glass curtain wall joint area is marked as the key scanning zone based on the structural drawings. When the UAV cruises to the middle level as instructed, the infrared sensor detects an abnormal temperature rise in a certain window frame area. The system immediately issues a reconnaissance command, and the UAV switches to reconnaissance mode to approach the area. The location was sampled intensively from multiple angles; the ground control center used an improved multispectral fusion algorithm, combined with the current floor wind speed and background thermal radiation level, to reduce the infrared mode weights while strengthening the confidence of the ultraviolet signal. The fusion results showed significant flame characteristics; after matching the PVC insulation material combustion template in the dynamic combustible material library, the fire was confirmed, and a standardized report containing coordinates and material type was generated. This report was pushed to the municipal fire command platform via the MQTT interface and marked as a medium-priority event based on personnel distribution information; after the mission was completed, the spectral characteristics of this smoldering process were stored in the decorative materials sub-library for reference in subsequent missions.

[0026] This invention achieves comprehensive coverage of fire-prone areas at different height levels through layered detection task planning, combined with the three-dimensional spatial structure characteristics of high-rise buildings. This solves the problem of insufficient coverage of special areas in high-rise buildings by traditional detection methods, significantly improving the comprehensiveness of detection. The close-range reconnaissance mode for abnormal areas enables precise data collection in key areas, enhancing the targeting of fire detection. The invention employs a judgment method combining multimodal data fusion and a dynamic combustible material database. The strong complementarity of multimodal data comprehensively reflects the state of the target area, while the dynamic combustible material database ensures adaptability to the combustion characteristics of various building materials, effectively reducing the false positive and false negative rates of single-data judgments and significantly improving the accuracy of fire judgment. Simultaneously, the application of data preprocessing and deep fusion algorithms further enhances the reliability of the judgment results.

[0027] In one embodiment of the present invention, the specific steps for generating task instructions are as follows: Step S101: The ground control center analyzes the received information about high-rise buildings, divides them into different height levels according to the building height, and clarifies the coverage of each level. The ground control center analyzes the height data of high-rise buildings and divides them into height levels using a "segmented adaptation" principle. This division is based on the layered management requirements for high-rise buildings in building fire protection design codes and the patterns of fire spread. In practice, the total building height is first extracted. Following the logic of "lower floors focusing on critical evacuation areas, middle floors focusing on structural connection risk areas, and upper floors focusing on concentrated equipment areas," the building is divided into three core levels, clearly defining the vertical height range and horizontal coverage area of ​​each level. For example, a building with a total height of 150 meters can be divided into 0-30 meter low floors, 30-80 meter middle floors, and 80-150 meter high floors. The coverage area of ​​each level is bounded by the building's exterior wall outline and includes all floors, balconies, equipment platforms, etc., within that height range, ensuring no blind spots. After the division is completed, the level information is stored in structured data format, providing a basic framework for subsequent path planning.

[0028] Step S102: Based on the building structure drawings, identify the fire-prone areas on each floor, and plan the detection path pattern according to the structural characteristics of different areas; By combining architectural structural drawings and using graphic recognition technology to extract key structural features of each floor, fire-prone areas can be accurately located. For lower floors, key areas to identify include doors and windows, balconies, entrance halls, and ground-floor shops. These areas have high pedestrian traffic and frequent electrical and fire-related activities, resulting in a higher fire risk. For mid-floor floors, key areas to identify include curtain wall joints, pipe shafts, equipment floors, and evacuation stairwells. These areas often have issues such as poor sealing, dense piping, and poor ventilation, making them prone to hidden fires. For upper floors, key areas to identify include rooftop equipment areas, elevator machine rooms, water tank rooms, and communication base stations. These areas have concentrated equipment, unique heat dissipation conditions, and a higher probability of electrical faults causing fires. Based on the structural characteristics of different fire-prone areas, corresponding detection path patterns are planned: For densely populated and open areas, a "fixed-point acquisition as the main method and cruise scanning as a supplementary method" mode is adopted. Fixed-point acquisition points are set at the center and four corners of the area to ensure full coverage of key point data, while the cruise scanning path circles around the edge of the area to supplement edge data acquisition; For areas with dense pipelines and narrow spaces, a "cruise scanning as the main method and fixed-point acquisition as a supplementary method" mode is adopted. The cruise scanning path is planned along the pipeline direction and spatial passage, while fixed-point acquisition points are set at key nodes where pipelines intersect and equipment is concentrated; For areas with complex structures and multiple obstructions, a "combination of fixed-point acquisition and cruise scanning" mode is adopted. Fixed-point acquisition points cover unobstructed key locations, while the cruise scanning path avoids obstructions and achieves full coverage by detouring around multiple angles.

[0029] Step S103: Referencing the characteristics of the building's surrounding environment and historical meteorological data, set appropriate sensor acquisition frequencies and UAV cruising speed parameters for each level. Based on the characteristics of the surrounding environment and historical meteorological data, a parameter adaptation model was established to set precise sensor acquisition frequencies and UAV cruising speeds for each level. Environmental characteristics mainly include the distribution of surrounding buildings, topography, and vegetation cover. Historical meteorological data focuses on extracting statistical data such as wind speed, wind direction, temperature, and humidity for the area over the past three years. For lower-level areas, where there is significant environmental interference (such as pedestrians, vehicles, and reflected light from surrounding buildings), a higher sensor acquisition frequency was set to ensure data validity, while the UAV cruising speed was moderate to allow sufficient time to filter interference signals. For mid-level areas, where the environment is relatively stable but significantly affected by airflow, a medium sensor acquisition frequency was set, and the cruising speed was dynamically adjusted according to airflow conditions, increasing when the airflow is stable and decreasing when it is complex. For upper-level areas, significantly affected by high-altitude wind speeds (typically higher than in lower and mid-level areas), a maximum sensor acquisition frequency was set to ensure rapid data capture of changes, while the cruising speed was appropriately reduced to ensure UAV flight stability and data acquisition accuracy. After the parameters are set, the task instruction template is embedded and associated with the hierarchical division and path planning data.

[0030] Step S104: Integrate the height hierarchy division results, detection path, acquisition frequency and cruise speed to generate standardized hierarchical detection mission instructions.

[0031] A standardized command generation framework is constructed, integrating information such as altitude level division results, detection path data, sensor acquisition frequency, and UAV cruise speed according to a preset format. First, the data of each component is standardized to ensure uniform data types and parameter units; for example, altitude data is in meters, acquisition frequency is in times per second, and cruise speed is in meters per second. Second, a data association mapping is established to clarify the detection path, acquisition frequency, and cruise speed corresponding to each altitude level, ensuring logical consistency of commands. Finally, a command encoding algorithm is used to convert the integrated data into standardized commands that the UAV can recognize and execute. These commands include core information such as mission identifiers, execution order, parameter configurations, and fault tolerance mechanisms. The fault tolerance mechanism is used to handle unexpected situations during flight (such as signal interruption or obstacle avoidance) to ensure stable mission execution. After command generation, a verification algorithm verifies its completeness and logic. Once confirmed to be error-free, it is stored in the ground control center command database, awaiting issuance.

[0032] This invention breaks through the limitations of traditional drones' "indiscriminate patrol" by planning dedicated paths according to building height and fire-prone areas. It concentrates detection resources on high-risk fire areas, achieving precise coverage of the three-dimensional space of high-rise buildings. This effectively avoids the problem of missing key areas or repeatedly detecting non-key areas, thus improving detection efficiency.

[0033] In one embodiment of the present invention, the specific steps of the preprocessing are as follows: Step S201: The edge computing unit on the UAV receives multimodal data and environmental parameters collected by the multimodal spectral detection unit and the environmental perception unit; Edge computing units are embedded computing modules deployed locally on the UAV. They have the ability to receive, process, store, and perform preliminary analysis of data. They can complete preprocessing operations at the data acquisition end without relying on the ground control center, reducing data transmission pressure and latency, and improving the system's real-time response performance.

[0034] The edge computing unit onboard the drone establishes real-time communication connections with the multimodal spectral detection unit and the environmental perception unit via hardware interfaces. The multimodal spectral detection unit includes an infrared sensor, an ultraviolet sensor, a thermal imager, a visible light camera, and a smoke sensor, simultaneously acquiring infrared intensity and wavelength, ultraviolet intensity and wavelength, temperature distribution images, visible light images of the scene, and smoke concentration data. The environmental perception unit includes a temperature and humidity sensor, a wind speed sensor, and a barometric pressure sensor, acquiring environmental temperature, humidity, wind speed, and barometric pressure data at the corresponding altitude levels. The edge computing unit receives the aforementioned multimodal data and environmental parameters according to a preset communication protocol, performs preliminary data classification and storage, and ensures the independence and integrity of different data types.

[0035] Step S202: Based on multimodal data and environmental parameters, corresponding preprocessing algorithms are used for noise filtering, signal enhancement, and data smoothing for different types of data. To address the characteristics of different data types, appropriate preprocessing algorithms are employed: For infrared and ultraviolet data, a median filtering algorithm is used to filter random noise. This algorithm effectively eliminates signal fluctuations caused by environmental interference by selecting the median of the data sequence to replace outliers. For visible light image data, a histogram equalization algorithm is used to enhance the signal. By adjusting the image grayscale distribution, the clarity of details is improved, facilitating subsequent identification of combustibles and location of abnormal areas. For thermal imaging data and smoke concentration data, a moving average algorithm is used for smoothing. By calculating the average value of a continuous data window, the impact of instantaneous fluctuations on data stability is eliminated, highlighting the trend changes related to fire conditions.

[0036] Step S203: Associate the preprocessed multimodal data with environmental parameters according to timestamps to form structured data; The edge computing unit extracts the acquisition timestamps of each set of multimodal data and environmental parameters, with timestamps accurate to the millisecond level to ensure data time synchronization. Using the timestamps as indexes, the preprocessed infrared data, ultraviolet data, thermal imaging data, visible light image data, and smoke concentration data are associated and bound with the corresponding temperature, humidity, wind speed, and air pressure parameters, forming a structured data set containing "spectral data - environmental parameters - time index." This clarifies the correspondence between data and provides a well-organized data foundation for subsequent multispectral fusion analysis.

[0037] Step S204: Compress the structured data and transmit it back to the ground control center in real time through the communication module to ensure the real-time performance and integrity of the data transmission.

[0038] A lightweight data compression algorithm is employed to compress structured data. This algorithm significantly reduces data size and bandwidth consumption while ensuring that data accuracy loss does not exceed 3%. The compressed structured data is transmitted to the ground control center in real-time via the 5G communication module on the UAV. A data verification mechanism is used during transmission to ensure that the data is not lost or tampered with during transmission, guaranteeing the integrity and reliability of the data received by the ground control center. This provides accurate data support for subsequent anomaly detection and fire assessment.

[0039] This invention effectively filters noise and interference signals in multimodal data through targeted preprocessing algorithms, enhances the identification of effective information, and solves the problems of excessive redundant information and poor signal stability in the original acquired data, providing high-quality data input for subsequent anomaly detection and multispectral fusion analysis.

[0040] In one embodiment of the present invention, the specific steps of transmitting the data to the fire command system via the fire alarm linkage interface are as follows: Step S501: The ground control center integrates the fire situation assessment results, the type of burning material, the fire level, and the priority ranking information of multiple fire points, and generates a standardized fire report according to the preset report format. The ground control center first invokes the data integration module to collect fire situation assessment results (including the conclusion of whether a fire has occurred), combustible material type (specific material names obtained by matching the dynamic combustible material database), fire severity level (levels classified based on burning area, peak temperature, and spread rate), and priority ranking information for multiple fire points (priority level and comprehensive score of each fire point). Then, following the pre-set report format of the national fire protection industry data standards, the above information is structurally integrated with collected environmental parameters (temperature, humidity, wind speed, air pressure), detection time (data collection start and end timestamps), and UAV positioning information (key coordinate points during the detection process). The logical relationships and presentation order of each field are clearly defined, ultimately generating a standardized fire report containing all core information without omission and with a unified format, ensuring that the report can be directly parsed and read by the fire command system.

[0041] Step S502: Transmit the fire report to the city fire command system through a standardized linkage interface compatible with general fire communication protocols to trigger dispatch and related emergency linkage commands; The standardized linkage interface adopts a combination of RS485 interface at the hardware level and TCP / IP protocol at the software level, pre-compatible with common fire communication protocols, including GB / T28181 video surveillance networking protocol and dedicated transmission protocol for fire protection IoT, achieving seamless integration with the urban fire command system. The ground control center establishes an encrypted communication link with the fire command system through this interface, employing a data fragmentation transmission and verification mechanism to ensure that fire reports are not lost or tampered with during transmission. Upon receiving a report, the fire command system automatically parses key information such as fire level, priority, and 3D coordinates, triggering preset emergency linkage logic: for level one priority fires, the highest level of dispatch is immediately initiated, simultaneously triggering commands such as elevator forced descent to the first floor, fire hydrant pressurization, directional broadcast of fire information via emergency broadcast, and closure of fireproof roller shutters to isolate the fire area; for level two and three priority fires, appropriate linkage measures are activated according to the corresponding level, ensuring a targeted and efficient response.

[0042] Step S503: The drone continues to patrol around the fire area, continuously collects dynamic data on the fire situation, and transmits it back to the ground control center in real time. After receiving cruise commands from the ground control center, the UAV adopts a fixed-altitude circling flight mode, maintaining a height range of 10 to 20 meters around the fire area and circling the fire area at a cruising speed of 3 meters per second. The multimodal spectral detection unit continuously collects data at a preset frequency. Among them, the infrared sensor and thermal imager collect temperature distribution and intensity data of the combustion area every 30 seconds, the visible light camera captures a set of high-definition images every minute, the smoke sensor collects smoke concentration data in real time, and the environmental sensing unit simultaneously records changes in environmental parameters such as temperature, humidity, and wind speed. All collected data is transmitted back to the ground control center in real time via a 5G high-speed communication network. During the transmission process, data compression and priority transmission strategies are used to ensure that key data (temperature changes, combustion area expansion) are transmitted first, ensuring the real-time nature of dynamic monitoring.

[0043] Step S504: The ground control center regularly updates the fire report based on the transmitted dynamic data, synchronizes it to the fire command system, and generates targeted prevention and control suggestions based on the type of combustible material.

[0044] The ground control center is equipped with a timed update mechanism, integrating and analyzing the transmitted dynamic data every minute. It compares the current data with the data from the previous period, including changes in burning area, peak temperature variations, and adjustments in spread rate. Based on this, it updates the corresponding fields in the fire report and synchronizes them to the city's fire command system, enabling commanders to monitor the fire's development in real time. Simultaneously, the ground control center accesses a prevention and control recommendation database. This database pre-stores the characteristics of various combustible materials and corresponding extinguishing methods and protective measures: for fires involving flammable and explosive materials, it generates recommendations such as "prioritize cutting off nearby power, using dry powder or carbon dioxide fire extinguishers, and strictly prohibiting direct water extinguishing"; for fires involving combustible solid materials, it generates recommendations such as "use water-based fire extinguishers or fire hoses, focusing on controlling the fire's spread path"; and for fires involving electrical equipment, it generates recommendations such as "first cut off power, use carbon dioxide fire extinguishers, and avoid the risk of electric shock," ensuring the professionalism and relevance of the prevention and control recommendations.

[0045] In one embodiment of the present invention, the specific process of collecting reconnaissance data is as follows: Step S301: The ground control center and the edge computing unit carried by the UAV establish a collaborative analysis mechanism to jointly determine anomalies in the preprocessed multimodal data; The ground control center deploys a data collaborative analysis system, establishing a real-time communication link with the edge computing unit carried by the UAV, forming a two-layer collaborative judgment architecture of "edge preprocessing + cloud verification". The edge computing unit first performs preliminary screening of the preprocessed multimodal data, removing obviously invalid data, and then uploads key feature parameters to the ground control center according to a preset data format. The ground control center simultaneously retrieves historical detection data and regional environmental baseline data, and cross-compares them with the data transmitted back by the UAV. The judgment criteria adopt a dual logic of "threshold triggering + trend analysis", presets the normal value range of each modality data, and monitors the changing trend of the data in continuous time series. When the data exceeds the threshold range or shows an abnormal changing trend, it is judged as an anomaly, ensuring the comprehensiveness and accuracy of anomaly detection.

[0046] Step S302: When any modal data reaches the preset anomaly judgment standard, the ground control center immediately issues a reconnaissance trigger command to the UAV; Once the collaborative judgment mechanism confirms an anomaly in any modal data, the command generation module at the ground control center is automatically activated. Based on key parameters such as the location information, timestamp, and modal type associated with the anomaly data, a standardized reconnaissance trigger command is generated. The command includes core information such as the preliminary location coordinates of the anomaly area, the reconnaissance distance range, sensor acquisition frequency adjustment parameters, and flight speed limits, and is encrypted and sent to the UAV via a low-latency communication protocol. To ensure the reliability of command transmission, a dual-link transmission mode of "main link + backup link" is adopted. The main link prioritizes the transmission of command data, while the backup link monitors the transmission status in real time. If the main link is interrupted, it immediately switches to the backup link to retransmit, ensuring that the UAV receives the reconnaissance command in a timely manner.

[0047] Step S303: After receiving the instruction, the UAV switches to reconnaissance mode and accurately locks the spatial coordinates of the abnormal area through the positioning module; The drone's positioning module employs a multi-source fusion positioning technology combining GPS, IMU, and visual positioning. GPS provides global position coordinates, the IMU collects real-time drone attitude data and motion parameters, and visual positioning matches environmental images captured by a visible light camera with a pre-set 3D building model to achieve local position correction. Upon receiving a reconnaissance command, the positioning module first analyzes the initial positioning coordinates in the command and quickly adjusts its flight attitude to approach the target area. During the approach, visual positioning technology compares real-time images with feature points on the building model, continuously correcting positional deviations, ultimately achieving centimeter-level precise locking of the spatial coordinates of the abnormal area, providing accurate positional support for subsequent close-range reconnaissance.

[0048] Step S304: Adjust the flight path, control the distance from the abnormal area, reduce the flight speed, increase the acquisition frequency of each sensor, and enable the visible light video recording function. After receiving reconnaissance instructions, the UAV's flight control system automatically switches to reconnaissance mode. Based on the coordinates of the locked anomaly area, it plans the optimal approach route, employing a "progressive approach" strategy to gradually shorten the distance to the anomaly area, eventually stabilizing within a preset close-range range. Simultaneously, the flight control system reduces its flight speed to a low-speed cruise state to ensure flight stability and prevent airflow disturbances from affecting data acquisition quality. The sensor control module responds synchronously, increasing the acquisition frequency of the corresponding modal sensor that triggered the anomaly by a preset ratio to ensure intensive acquisition of key data; the acquisition frequencies of other modal sensors are also adjusted synchronously to ensure the integrity of multi-dimensional data; after the visible light video recording function is activated, it uses high-definition recording mode to record the visual characteristics of the anomaly area, providing intuitive evidence for subsequent fire assessment.

[0049] Step S305: Continuously collect multimodal data and environmental parameters, associate them according to preset rules to form a complete reconnaissance data packet, and transmit it back to the ground control center in real time through a high-speed communication network.

[0050] During close-range reconnaissance, the multimodal spectral detection unit continuously collects multi-dimensional data including infrared, ultraviolet, thermal imaging, visible light, and smoke. The environmental perception unit simultaneously records environmental parameters such as temperature, humidity, wind speed, and air pressure. All data is accompanied by precise timestamps and location coordinates. The edge computing unit receives the collected data in real time and, according to the principle of "timestamp alignment + location coordinate association," structurally integrates the multimodal data with environmental parameters to form a reconnaissance data packet containing information such as data type, collection time, collection location, and environmental conditions. During the transmission process, an adaptive transmission mechanism is used, dynamically adjusting the data transmission rate based on the communication network bandwidth, prioritizing the transmission of key modal data, and simultaneously performing block compression processing on the data packet to reduce transmission latency. The data is then transmitted back to the ground control center in real time and completely through a high-speed communication network, ensuring that the ground end can promptly obtain high-quality reconnaissance data for fire situation assessment.

[0051] This invention utilizes a collaborative judgment mechanism between the ground control center and the edge computing unit, combined with the dual logic of "threshold triggering + trend analysis," to effectively avoid false detections or missed detections caused by a single judgment dimension. This significantly improves the accuracy and reliability of anomaly detection, ensuring that early fires or potential risks are not overlooked.

[0052] In one embodiment of the present invention, the specific steps for processing reconnaissance data using the improved multispectral fusion algorithm are as follows: Step S401: After receiving the reconnaissance data packet transmitted back by the UAV, the ground control center extracts the environmental parameters, including temperature, humidity, wind speed, and air pressure data. After receiving the reconnaissance data packets transmitted back by the UAV via a high-speed communication interface, the ground control center initiates a data parsing program. This program first unpacks the reconnaissance data packets, separating multimodal spectral data and environmental parameter data. The environmental parameters specifically focus on three core data types: temperature and humidity, wind speed, and air pressure. Temperature and humidity data are directly extracted from the results collected by the temperature and humidity sensors in the environmental sensing unit; wind speed data comes from the real-time monitoring records of the wind speed sensor; and air pressure data is taken from the detection output of the air pressure sensor. All three types of data retain their original acquisition accuracy without additional interpolation processing, ensuring the authenticity of the foundational data for subsequent fusion calculations.

[0053] Step S402: Based on the height and hierarchical characteristics of high-rise buildings, determine the weighting factors of each spectral mode of infrared, ultraviolet, thermal imaging, visible light, and smoke, and dynamically adjust the confidence level of each mode according to temperature and humidity parameters. Based on the hierarchical characteristics of high-rise buildings, a hierarchical adaptation weight allocation strategy is adopted. High-rise areas have strong air circulation and complex environmental interference factors; infrared and thermal imaging data are relatively less affected by the environment and are more sensitive to fire conditions. Therefore, the weight factor for the infrared spectral mode is set to 0.35, and the weight factor for the thermal imaging spectral mode is set to 0.3. Mid-level areas have relatively stable environmental conditions, and the effectiveness of each spectral mode data is balanced; therefore, the weight factors for the infrared, ultraviolet, thermal imaging, visible light, and smoke spectral modes are set to 0.3, 0.2, 0.25, 0.1, and 0.15, respectively. Low-rise areas are easily affected by ground light sources and pedestrian activity; ultraviolet and smoke data are more targeted for early fire identification. Therefore, the weight factor for the ultraviolet spectral mode is set to 0.25, the weight factor for the smoke spectral mode is set to 0.2, and the weight factors for the infrared, thermal imaging, and visible light spectral modes are set to 0.25, 0.2, and 0.1, respectively.

[0054] Modal confidence level adjustment employs an environmental parameter linkage mechanism. When the ambient temperature is above 35℃ and the relative humidity is above 60%, ultraviolet light is easily scattered by water vapor, leading to a decrease in data reliability. Therefore, the confidence level of the ultraviolet mode is reduced by 30%. Meanwhile, thermal imaging data can still stably reflect temperature distribution under high temperature and high humidity environments, so the confidence level of the thermal imaging mode is increased by 20%. When the ambient temperature is below 5℃ and the relative humidity is below 30%, infrared data is more sensitive to fire radiation in low-temperature environments. Therefore, the confidence level of the infrared mode is increased by 25%. Smoke diffuses rapidly in low humidity environments, resulting in insufficient stability of concentration data. Therefore, the confidence level of the smoke mode is reduced by 15%. Under other environmental conditions, the confidence levels of each mode remain unchanged from the baseline value.

[0055] Step S403: Use a weighted fusion method to integrate and calculate the multimodal data in the reconnaissance data packet, eliminate the limitations of single-modal data, and generate comprehensive fire feature values; A weighted fusion algorithm based on confidence correction is employed. First, the raw data for each spectral mode are standardized to eliminate the influence of differences in the dimensions of data from different sensors. Then, the standardized modal data are multiplied by their corresponding weighting factors and modal confidence scores to obtain corrected data for each mode. Finally, the corrected data for all modes are summed to generate a comprehensive fire characteristic value. This value encompasses multi-dimensional information such as infrared intensity and wavelength, ultraviolet intensity and wavelength, thermal imaging temperature distribution, visible light image characteristics, and smoke concentration, effectively eliminating the limitations of single-modal data and improving the comprehensiveness of fire characteristics.

[0056] Step S404: Compare the comprehensive fire characteristic value with the combustion characteristic data of various materials in the dedicated dynamic combustible material warehouse for high-rise buildings one by one; A feature value-by-dimensional comparison mechanism is established to compare the generated comprehensive fire feature value with the combustion feature data of various materials in the dynamic combustible material database specifically for high-rise buildings. During the comparison process, similarity calculations are performed on multiple dimensions, including infrared related features, ultraviolet related features, thermal imaging related features, and smoke concentration features. The similarity calculation for each dimension adopts the Euclidean distance algorithm. By calculating the distance values ​​between the comprehensive fire feature value and the material combustion feature data in the database in each dimension, the degree of fit between the two is judged. The smaller the distance value, the higher the similarity.

[0057] Step S405: Determine whether a fire is established based on the comparison results. If a fire is established, further determine the corresponding type of combustible material and determine the fire level by combining relevant combustion characteristic indicators.

[0058] A similarity threshold is set as the basis for fire severity determination. When the comprehensive fire feature value has a multi-dimensional average similarity of more than 85% with the combustion feature data of a certain type of material in the database, a fire is determined, and the material is identified as the corresponding combustible material type. Simultaneously, fire severity levels are classified based on combustion-related feature indicators. Combustion area, peak temperature, and spread rate are used as core level determination indicators. The combustion area is calculated using visible light image segmentation and pixel counting, the peak temperature is taken from the highest temperature point in thermal imaging data, and the spread rate is calculated from the boundary changes of the combustion area in consecutive frame images. Based on the specific values ​​of these three indicators, fires are divided into three levels, achieving precise quantification of fire severity.

[0059] This invention dynamically adjusts weight factors and modal confidence by combining height-level characteristics and environmental parameters, enabling the fusion algorithm to adapt to the environmental differences in different areas of high-rise buildings. This significantly improves the targeting and reliability of fire feature extraction in complex environments, and the accuracy of fire determination is improved by more than 30% compared with traditional fixed-weight fusion algorithms.

[0060] In one embodiment of the present invention, the construction and updating of the dedicated dynamic combustible material storage for high-rise buildings is performed according to the following steps: Step S406: Collect common materials for high-rise buildings, covering building structural materials, decorative materials, pipe materials, and electrical equipment-related materials; determine the spectral characteristics and combustion-related parameters of various materials under different environmental conditions; and construct a basic combustible material library. By surveying urban high-rise building construction standards, mainstream building materials market product lists, and building industry material application reports, we systematically collected common materials, specifically covering building structural materials such as reinforced concrete and steel structures, decorative materials such as wood panels and plastic decorative parts, pipe materials such as PVC and stainless steel, and electrical equipment related materials such as cable insulation and electrical housings, ensuring coverage of the core material categories used in high-rise buildings.

[0061] In a standard laboratory environment, combustion experiments were conducted on various materials under simulated temperature, humidity, and air pressure conditions (covering common environmental ranges for low-rise, mid-rise, and high-rise buildings). A high-precision spectrometer was used to collect infrared and ultraviolet spectral characteristics during the combustion process. Temperature sensors and smoke concentration detectors were used to record combustion-related parameters such as changes in combustion temperature and smoke release. All data was categorized and stored according to material type and environmental conditions, constructing a clearly structured basic combustible material library.

[0062] Step S407: During the detection process, when the spectral data collected by the UAV cannot match the data in the basic combustible material library, the edge computing unit extracts the spectral feature value of the unknown material and associates it with the environmental parameters and visible light image features at the time of collection. When the UAV is performing a detection mission, the edge computing unit compares the spectral data collected by the multimodal spectral detection unit with the data in the basic combustible material library in real time, and uses the feature value similarity algorithm to determine whether they match. When the similarity is lower than the preset threshold, it is determined to be an unknown material.

[0063] The edge computing unit automatically extracts the spectral feature values ​​of the unknown material, including core parameters such as infrared intensity and wavelength distribution, ultraviolet intensity and wavelength distribution. At the same time, it associates environmental parameters such as temperature, humidity, wind speed, and air pressure acquired by the environmental sensing unit at that moment, as well as image features such as the appearance and shape of the material captured by the visible light camera, to form a complete feature dataset of the unknown material.

[0064] Step S408: The extracted data related to unknown materials is uploaded to the ground control center via the communication network, and the ground control center synchronizes it to the city fire protection big data platform; The extracted dataset of unknown material features is encrypted and uploaded to the ground control center via the 5G communication module on the drone. During the transmission process, a data fragmentation and verification mechanism is used to ensure that the data is complete and without loss.

[0065] After receiving the data, the ground control center synchronizes it to the city's fire protection big data platform through a standardized data interface. During the synchronization process, metadata such as data collection time, drone number, and detection location are automatically recorded to facilitate subsequent traceability.

[0066] Step S409: The platform assigns professional reviewers to review the data, make a preliminary judgment on the material properties, and start a simulated combustion experiment after the review is passed to determine the complete combustion characteristic parameters of the material under different environmental conditions. The city's fire protection big data platform assigns data to three or more auditors with relevant professional qualifications in building materials and fire protection engineering, based on rules such as material type and collection area. The auditors use the platform to view the material's spectral curves, image characteristics, and collection environment information, and independently make a preliminary judgment on the material's properties to form a review opinion.

[0067] When more than half of the reviewers agree, a simulated combustion experiment is initiated. The experiment is conducted in a controlled environment chamber, accurately replicating the environmental conditions under which the unknown material was collected. Multi-dimensional detection equipment is used to comprehensively measure complete combustion characteristic parameters such as spectral characteristics, temperature changes, smoke composition, and combustion rate.

[0068] Step S410: After the experimental data is verified to be correct, it is submitted to the dynamic combustible material storage management system, and the system automatically updates the data in the storage.

[0069] After the experiment is completed, two or more professional laboratory personnel cross-validate the experimental data to verify the accuracy and completeness of the data and ensure that there are no measurement errors or experimental deviations.

[0070] Once the combustion characteristic parameters are verified, they are submitted to the dynamic combustible material library management system. The system automatically enters the material information into the library, establishes a correlation mapping between material properties, environmental conditions, and combustion characteristics, and completes the dynamic expansion of the basic combustible material library.

[0071] This invention uses a dynamic update mechanism to promptly incorporate the combustion characteristics of new and special materials in high-rise buildings into the combustible material database, solving the problem of judgment errors caused by the inability of traditional static databases to cover new materials, and significantly improving the accuracy of identifying the combustion of various materials.

[0072] See Figure 2 In one embodiment of the present invention, when multiple fire points are identified, it is necessary to calculate the priority ranking of each fire point. The specific steps are as follows: Step S601: The ground control center collects the positioning data and multi-frame visible light image data transmitted back by the UAV; The ground control center receives positioning data and multiple frames of visible light imagery data transmitted from the UAV via a wireless communication link. The positioning data, generated by the UAV's GPS / IMU integrated navigation module, includes latitude, longitude, altitude, and attitude angle information at the time of each image acquisition, achieving centimeter-level positioning accuracy to ensure precise spatial correlation of the fire location. The visible light imagery data is acquired by a high-definition industrial camera onboard the UAV, with a resolution of at least 4K and a frame rate of 15 to 30 frames per second, ensuring clear image details and good temporal continuity, covering the complete scene of the fire location and surrounding area. The ground control center performs integrity verification on the received data, eliminating data lost or distorted during transmission to ensure the reliability of subsequent processing.

[0073] Step S602: Use a 3D modeling algorithm to process the positioning data and multi-frame visible light image data to generate a complete 3D model of the high-rise building. The SFM (Structured Motion Recovery) algorithm was used to process the collected positioning data and multi-frame visible light imagery. First, feature points were extracted and matched from the multi-frame visible light images. The SIFT algorithm was used to identify corresponding feature points between images, and the spatial geometric relationships between these feature points were calculated. Combining the attitude angles and position information from the positioning data, constraints between images were constructed. The bundle adjustment algorithm was used to optimize the spatial point coordinates and camera parameters, generating a sparse point cloud. Dense reconstruction was then performed based on the sparse point cloud, and the gaps in the point cloud were filled using the Poisson reconstruction algorithm to form a high-density point cloud model. Finally, mesh generation and texture mapping were performed, attaching the texture information from the visible light images to the surface of the mesh model to generate a complete and realistic 3D model of a high-rise building. The model accuracy meets the requirement that the fire location error does not exceed 0.5 meters.

[0074] Step S603: Accurately mark the spatial location, burning range, and spread information of each fire point in the three-dimensional model of the high-rise building; Based on the generated 3D model of the high-rise building, and combined with thermal imaging temperature distribution and smoke concentration distribution information from multimodal data, the spatial location of each fire point is accurately located. A threshold segmentation algorithm is used to extract temperature anomaly areas from the thermal imaging data, determining the core range of the fire points, and mapping them to the corresponding latitude, longitude, and altitude coordinates of the 3D model. Image segmentation and edge detection algorithms are used to process visible light images to obtain the combustion boundaries of the fire points, and the dynamic changes in the combustion range are calculated by combining time-series image data. By comparing fire areas in adjacent frames of images, the direction and speed of flame spread are analyzed. Information such as spatial location, combustion range, spread direction, and spread speed are superimposed onto the 3D model in the form of visual markers, intuitively presenting the real-time status of each fire point.

[0075] Step S604: Establish an assessment model that includes three dimensions: personnel density, combustible material hazard, and spread risk, and clarify the evaluation criteria for each dimension; An assessment model was established, encompassing three dimensions: personnel density, combustible material hazard, and spread risk. The evaluation criteria and quantification methods for each dimension were clearly defined. Personnel density dimension: Based on the structural drawings and personnel distribution information of the high-rise building, the personnel carrying capacity standards for each area within the building were determined. The actual personnel density within each area was judged by combining real-time detection data and quantified as a value from 0 to 1, with higher values ​​indicating greater personnel density. Combustible material hazard dimension: Referring to the combustion characteristics of various materials in the high-rise building's dedicated dynamic combustible material database, combustibles were classified into three hazard levels: high, medium, and low, corresponding to quantitative scores of 0.8, 0.4, and 0.1 respectively. Higher hazard levels resulted in higher scores. Spread risk dimension: Based on the spread speed of the fire, the density of surrounding combustible materials, and the building's ventilation conditions, a value from 0 to 1 was quantified. Faster spread speed, denser surrounding combustible materials, and better ventilation conditions resulted in higher scores.

[0076] Step S605: Input the relevant data of each fire point into the evaluation model, calculate the comprehensive score, and divide different priority levels according to the score results.

[0077] The quantitative values ​​of population density, combustible material hazard, and spread risk at each fire location are input into the evaluation model. A weighted summation method is used to calculate the comprehensive score. The weight allocation is set according to the urban fire rescue priority criteria: population density weight is 0.4, combustible material hazard weight is 0.3, and spread risk weight is 0.3. The comprehensive score = population density quantitative value × 0.4 + combustible material hazard score × 0.3 + spread risk quantitative value × 0.3. Based on the comprehensive score results, three priority levels are divided: a comprehensive score of not less than 0.7 is the first priority level, corresponding to densely populated areas, high-risk combustible material areas, or fire locations with high spread risk; a comprehensive score between 0.4 and 0.7 is the second priority level, corresponding to medium-density areas, medium-risk combustible material areas, or fire locations with medium spread risk; and a comprehensive score below 0.4 is the third priority level, corresponding to low-density areas, low-risk combustible material areas, or fire locations with low spread risk.

[0078] This invention uses a three-dimensional model to intuitively present the spatial distribution, burning range, and spread trend of each fire point, solving the problem that traditional two-dimensional images cannot accurately reflect the three-dimensional fire situation in high-rise buildings, and providing fire rescue personnel with a clear global view of the fire situation.

[0079] In one embodiment of the present invention, the UAV is equipped with a multimodal spectral detection unit and an environmental perception unit, and has automatic calibration and low-power coordination, as detailed below: Step S701: Before each reconnaissance mission, the UAV starts a self-test calibration program at the ground take-off and landing point and calls the preset standard reference object; Before performing a detection mission, the UAV must complete a self-test calibration at the ground take-off and landing point to ensure the basic accuracy of the collected data. First, the ground control center sends a self-test calibration command to the UAV. Upon receiving the command, the UAV's flight control system activates its built-in self-test calibration program. This program establishes a communication connection with the multimodal spectral detection unit and the environmental perception unit, synchronously calling up preset standard references. The standard references include a standard spectral plate and a standard environmental parameter source. The standard spectral plate integrates infrared and ultraviolet spectral characteristics with known fixed intensities and wavelengths. The standard environmental parameter source provides stable reference values ​​for temperature, humidity, wind speed, and air pressure. All standard parameters comply with national fire detection equipment calibration specifications, providing a unified reference standard for sensor calibration.

[0080] Step S702: The multimodal spectral detection unit and the environmental sensing unit respectively collect relevant data of the standard reference object and compare them with the preset standard data; The infrared sensor, ultraviolet sensor, thermal imager, visible light camera, and smoke sensor in the multimodal spectral detection unit collect the corresponding spectral characteristics of the standard spectral plate and the relevant environmental parameters of the standard environmental parameter source, respectively, to obtain standard acquisition data. The temperature and humidity sensor, wind speed sensor, and air pressure sensor in the environmental perception unit simultaneously collect the reference data of the standard environmental parameter source. After the acquisition is completed, the edge computing unit obtains the acquired data from each sensor and retrieves the standard data pre-stored in the UAV's local storage module. Through point-to-point comparison, it calculates the deviation value between the data acquired by each sensor and the corresponding standard data, forming a deviation value matrix to ensure that the acquisition accuracy of each sensor can be independently evaluated.

[0081] Step S703: Calculate the deviation between the data collected by various sensors and the standard data. When the deviation exceeds the allowable range, automatically adjust the detection parameters of the sensors. The edge computing unit determines whether the deviation of each sensor exceeds the preset allowable range based on the deviation value matrix. For infrared sensors, the allowable deviation range is set to be no more than 5% of the difference between the collected data and the standard data. When this range is exceeded, the sensor's spectral response gain parameter is automatically adjusted, and the signal conversion coefficient is optimized until the deviation value falls back to the allowable range. For ultraviolet sensors, the allowable deviation range is no more than 8% of the difference of the standard data. When the deviation exceeds the limit, the spectral receiving sensitivity is corrected by adjusting the sensor's detection threshold voltage. For thermal imagers, the allowable deviation range is no more than 3℃ of temperature difference. When the deviation exceeds the limit, the temperature mapping curve of the thermal imaging pixels is calibrated to improve temperature measurement accuracy. For temperature and humidity sensors, wind speed sensors, and barometric pressure sensors, the allowable deviation ranges are ±2℃, ±0.3 m / s, and ±5 kPa, respectively. When the deviation exceeds the limit, the hardware compensation coefficient of the sensors is corrected to achieve parameter calibration and ensure the accuracy of the data collected by all sensors.

[0082] Step S704: After calibration, the UAV takes off to perform the detection mission. If no abnormal data is detected, switch to low power operation mode and adjust the cruise speed and sensor acquisition frequency. After calibration, the UAV flight control system receives takeoff instructions from the ground control center and activates the power system to execute the detection mission. During the anomaly detection phase, the flight control system triggers a low-power coordination mechanism, adjusting the UAV's cruising speed from the normal 3-5 m / s to 2 m / s to reduce power system energy consumption. Simultaneously, the edge computing unit sends low-power commands to the multimodal spectral detection unit and the environmental perception unit, adjusting the sensor acquisition frequency from the normal 1-2 times / s to 1 time / 5 seconds, retaining only core detection functions and disabling unnecessary data preprocessing algorithms. The edge computing unit itself switches to a low-power operating state, reducing the processor's clock speed and power consumption. Through multi-dimensional parameter adjustments, the system maximizes the UAV's endurance for a single detection mission.

[0083] Step S705: When abnormal data is detected, automatically exit the low power mode and restore normal operating parameters to ensure the accuracy of abnormal area detection.

[0084] During low-power detection, the edge computing unit continuously monitors the multimodal data collected by the sensors in real time. When any modal data reaches the preset anomaly judgment standard, it immediately sends a wake-up command to the flight control system and each detection unit. The flight control system quickly restores the cruise speed to the preset value of the normal mode, ensuring that the UAV can respond quickly to the abnormal area; the multimodal spectral detection unit and the environmental perception unit stop low-power operation and restore the normal acquisition frequency and all detection functions; the edge computing unit increases the processor's main frequency and starts the complete data preprocessing and analysis algorithm to ensure that the detection data of the abnormal area can be processed accurately and quickly, providing high-quality data support for subsequent fire situation determination.

[0085] This invention effectively corrects detection deviations caused by long-term use and environmental changes in sensors through an automatic calibration process before the mission, ensuring the accuracy of multimodal data and environmental parameter acquisition. This provides a reliable data foundation for subsequent multispectral fusion analysis and fire situation determination, reducing the risk of misjudgment or missed judgment of fire situation due to data errors.

[0086] In one embodiment of the present invention, the update review of the dynamic combustible material inventory is performed according to the following steps: Step S411: The urban fire protection big data platform receives unknown material-related data synchronized from the ground control center, including spectral characteristic values, acquisition environment parameters, and visible light image characteristics. The city's fire protection big data platform receives data related to unknown materials synchronously transmitted from the ground control center through a pre-set standardized data interface. Data transmission employs an encrypted transmission protocol to ensure data security and integrity during transmission. The platform has a built-in data classification module that automatically categorizes and organizes the received data. Spectral feature values ​​are broken down and categorized according to the intensity and wavelength parameters corresponding to infrared, ultraviolet, and thermal imaging. Environmental parameters are archived separately according to temperature, humidity, wind speed, and air pressure. Visible light image features are labeled according to dimensions such as shooting angle, shooting distance, and image clarity, forming a structured data set that provides clear data support for subsequent review work.

[0087] Step S412: The platform distributes the data to qualified auditors according to preset rules, and the auditors independently review and judge the material attributes. Based on preset task allocation rules, and considering the reviewers' professional fields, workload, and historical review accuracy, the platform uses an intelligent allocation algorithm to assign unknown material data to three or more reviewers with relevant professional qualifications such as fire protection engineering and materials science. Reviewers log into the system through the platform's dedicated review terminal to independently view the unknown material's collection scene description, complete spectral curve, multi-angle visible light images, and corresponding environmental parameters. Using the platform's built-in material feature comparison tool, they conduct a preliminary comparison of the unknown material's appearance and spectral characteristics with a known material database. Combining their professional knowledge, they determine the material's properties and form an independent review opinion. This review opinion must clearly indicate the preliminary material determination type, the basis for the determination, and any uncertainties.

[0088] Step S413: Based on the review results of multiple auditors, determine whether to initiate a simulated combustion experiment to measure the combustion characteristic parameters of the material; The platform's data processing module automatically aggregates the review opinions of all reviewers and calculates the consistency rate. When the consistency rate reaches a preset threshold (e.g., 80% or higher) and all review opinions clearly define the material attribute category, the platform directly determines whether to initiate a simulated combustion experiment. When the consistency rate is lower than the preset threshold, or when there are significant disagreements in the review opinions, the platform reassigns the data to two senior reviewers for a second review. The senior reviewers, combining the initial review opinions with supplementary relevant industry information, provide a final review conclusion to determine whether to initiate a simulated combustion experiment. If a consensus cannot be reached after the second review, the data is archived in a research database for further evaluation after accumulating more relevant data.

[0089] Step S414: After the experiment is completed, professionals will verify the experimental data and generate a complete report on the material combustion characteristics. The simulated combustion experiment was conducted in a fire science laboratory that meets national standards. The experimental setup included a combustion furnace, multispectral data acquisition equipment, environmental parameter control equipment, and data recording equipment. Before the experiment, different combinations of environmental conditions were set based on the material properties determined by the reviewers, covering common temperature, humidity, and air pressure ranges. The unknown material was processed into a standard sample and placed in the combustion furnace for simulated combustion. Infrared, ultraviolet, and thermal imaging spectral characteristic data were collected in real time during the combustion process using the multispectral data acquisition equipment, while environmental parameters and changes in the combustion state were recorded simultaneously. After the experiment, two professional researchers independently processed and analyzed the experimental data, verifying its accuracy and completeness, removing outlier data, supplementing missing data, and generating a complete combustion characteristic report containing information such as the full-cycle spectral characteristics of the material's combustion, temperature change curves, and smoke release parameters.

[0090] Step S415: Enter the verified combustion characteristic parameters into the dynamic combustible material database, record the update time, reviewer, experimenter and other relevant information, and form an update log.

[0091] Verified combustion characteristic reports are submitted to the dynamic combustible material database management system. The system uses an incremental update method, entering new material combustion characteristic parameters into the database according to a preset data structure and establishing a correlation index with existing data to ensure efficient subsequent queries and comparisons. Simultaneously, the system automatically records detailed information for each update, including update time, source of original data for unknown materials, name and qualification number of the reviewer, name of the experimenter, model of the experimental apparatus, experimental environmental conditions, and details of the entered combustion characteristic parameters, forming an immutable update log. The log data is backed up regularly and can be retrieved and viewed at any time, providing a basis for data traceability, responsibility determination, and subsequent algorithm optimization.

[0092] This invention employs a dual verification mechanism of "multiple rounds of independent review + professional experimental verification" to effectively filter out erroneous data and subjective judgment biases. This ensures the accuracy and reliability of the combustion characteristic parameters of novel materials included in the dynamic combustible material database, providing precise data support for fire assessment and reducing the risk of misjudgment due to inaccurate material data. A standardized process for collecting, reviewing, experimenting with, and storing unknown material data is established, enabling rapid inclusion of novel structural materials, decorative materials, and other materials with unknown combustion characteristics in high-rise buildings. This breaks through the adaptation limitations of traditional static combustible material databases, allowing the fire detection system to respond promptly to iterative changes in materials and expand its detection coverage.

[0093] The above description is only a preferred embodiment of the present invention. It should be noted that those skilled in the art can make several improvements and modifications without departing from the principle of the present invention, and these improvements and modifications should also be considered within the scope of protection of the present invention.

Claims

1. A method for detecting urban high-rise building fires using a multispectral fusion algorithm by unmanned aerial vehicles (UAVs), characterized in that, Includes the following steps: The ground control center receives the address, height, structural drawings, and personnel distribution information of high-rise buildings, converts them into layered detection parameters, generates mission instructions, and sends them to the UAVs equipped with multimodal spectral detection units. The UAV collects multimodal data in layers according to mission instructions, and transmits the pre-processed data back to the ground control center in real time. When the ground control center detects any abnormality in any modal data, it issues a reconnaissance command, and the UAV switches to reconnaissance mode to collect reconnaissance data at close range and transmit it back. The ground control center processes reconnaissance data using an improved multispectral fusion algorithm, and combines it with a pre-set dynamic combustible material warehouse specific to high-rise buildings to complete the fire situation assessment and generate the assessment result; A standardized fire report is generated based on the judgment results and transmitted to the fire command system through the fire linkage interface. The dynamic combustible material inventory is updated after the task is completed.

2. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, The specific steps for generating task instructions are as follows: The ground control center analyzes the received information about high-rise buildings, divides them into different height levels according to their height, and clarifies the coverage area of ​​each level. Based on the building structure drawings, identify fire-prone areas on each floor, and plan detection path patterns according to the structural characteristics of different areas; Based on the characteristics of the building's surrounding environment and historical meteorological data, appropriate sensor acquisition frequencies and drone cruising speed parameters are set for each level. By integrating the results of the hierarchical classification, the detection path, the acquisition frequency, and the cruise speed, standardized hierarchical detection mission instructions are generated.

3. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, The specific steps of the preprocessing are as follows: The edge computing unit onboard the drone receives multimodal data and environmental parameters collected by the multimodal spectral detection unit and the environmental perception unit; Based on multimodal data and environmental parameters, corresponding preprocessing algorithms are used for noise filtering, signal enhancement, and data smoothing for different types of data. The preprocessed multimodal data is associated with environmental parameters by timestamp to form structured data; Structured data is compressed and transmitted back to the ground control center in real time via a communication module to ensure the real-time performance and integrity of data transmission.

4. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, The specific steps for transmitting data to the fire command system via the fire alarm linkage interface are as follows: The ground control center integrates fire situation assessment results, type of burning material, fire level, and priority ranking information of multiple fire points, and generates a standardized fire report according to the preset report format. Through a standardized linkage interface compatible with general fire communication protocols, fire reports are transmitted to the city's fire command system, triggering dispatch and related emergency linkage commands; The drones maintained a patrol around the fire area, continuously collecting data on the dynamic changes in the fire situation and transmitting it back to the ground control center in real time. Based on the dynamic data transmitted back, the ground control center regularly updates the fire report, synchronizes it with the fire command system, and generates targeted prevention and control recommendations according to the type of combustible material.

5. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, The specific process for collecting reconnaissance data is as follows: The ground control center and the edge computing unit carried by the UAV establish a collaborative analysis mechanism to jointly determine anomalies in the preprocessed multimodal data; When any modal data reaches the preset anomaly judgment standard, the ground control center immediately issues a reconnaissance trigger command to the UAV; After receiving instructions, the drone switches to reconnaissance mode and uses the positioning module to accurately locate the spatial coordinates of the abnormal area. Adjust the flight path, control the distance from the abnormal area, reduce the flight speed, increase the acquisition frequency of each sensor, and enable the visible light video recording function. It continuously collects multimodal data and environmental parameters, associates them according to preset rules to form a complete reconnaissance data packet, and transmits it back to the ground control center in real time through a high-speed communication network.

6. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, The specific steps for processing reconnaissance data using the improved multispectral fusion algorithm are as follows: After receiving the reconnaissance data packets transmitted back by the UAV, the ground control center extracts the environmental parameters, including temperature, humidity, wind speed, and air pressure data. Based on the height and hierarchical characteristics of high-rise buildings, the weighting factors of each spectral mode, including infrared, ultraviolet, thermal imaging, visible light, and smoke, are determined, and the confidence level of each mode is dynamically adjusted according to temperature and humidity parameters. A weighted fusion method is used to integrate and calculate the multimodal data in the reconnaissance data packet, eliminating the limitations of single-modal data and generating comprehensive fire feature values; The comprehensive fire characteristic values ​​were compared one by one with the combustion characteristic data of various materials in the dedicated dynamic combustible material warehouse of high-rise buildings; The fire situation is determined based on the comparison results. If the fire situation is confirmed, the type of combustible material is further identified, and the fire level is determined in combination with relevant combustion characteristic indicators.

7. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, The construction and updating of the dedicated dynamic combustible material storage facility for high-rise buildings shall be carried out in accordance with the following steps: Collect common materials for high-rise buildings, covering building structural materials, decorative materials, pipe materials, and electrical equipment-related materials; determine the spectral characteristics and combustion-related parameters of various materials under different environmental conditions; and construct a basic combustible material library. When the spectral data collected by the UAV during the exploration process cannot be matched with the data in the basic combustible material database, the edge computing unit extracts the spectral feature value of the unknown material and associates it with the environmental parameters and visible light image features at the time of collection. The extracted data related to unknown materials is uploaded to the ground control center via a communication network, and then synchronized to the city's fire protection big data platform. The platform assigns professional reviewers to review the data, make a preliminary judgment on the material properties, and after the review is passed, start a simulated combustion experiment to determine the complete combustion characteristic parameters of the material under different environmental conditions. After the experimental data is verified to be correct, it is submitted to the dynamic combustible material storage management system, and the system automatically updates the data in the storage.

8. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, When multiple fire points are identified, it is necessary to calculate the priority ranking of each fire point. The specific steps are as follows: The ground control center collects the positioning data and multiple frames of visible light image data transmitted back by the UAV; A 3D modeling algorithm is used to process the positioning data and multi-frame visible light image data to generate a complete 3D model of the high-rise building. Accurately mark the spatial location, burning range, and spread information of each fire point in the 3D model of the high-rise building; Establish an assessment model that includes three dimensions: personnel density, combustible material hazard, and spread risk, and clarify the evaluation criteria for each dimension; The relevant data of each fire point are input into the evaluation model to calculate the comprehensive score, and different priority levels are divided according to the score results.

9. The UAV method for detecting urban high-rise building fires using a multispectral fusion algorithm according to claim 1, characterized in that, The UAV is equipped with a multimodal spectral detection unit and an environmental perception unit, and features automatic calibration and low-power coordination, as detailed below: Before each reconnaissance mission, the drone initiates a self-check and calibration procedure at the ground take-off and landing point, calling up preset standard reference objects; The multimodal spectral detection unit and the environmental sensing unit respectively collect relevant data from the standard reference object and compare them with the preset standard data; Calculate the deviation between the data collected by various sensors and the standard data, and automatically adjust the sensor's detection parameters when the deviation exceeds the allowable range; After calibration, the drone takes off to perform a detection mission. If no abnormal data is detected, it switches to a low-power operation mode and adjusts the cruise speed and sensor acquisition frequency. When abnormal data is detected, the system automatically exits low-power mode and restores normal operating parameters to ensure the accuracy of abnormal area detection.

10. The UAV method for detecting urban high-rise building fires using the multispectral fusion algorithm according to claim 7, characterized in that, The update review of the dynamic combustible material inventory is performed according to the following steps: The urban fire protection big data platform receives data related to unknown materials synchronously from the ground control center, including spectral characteristic values, acquisition environment parameters, and visible light image characteristics. The platform distributes data to qualified auditors according to preset rules, and the auditors independently review and judge the attributes of the materials. Based on the review results of multiple auditors, it was determined whether to initiate a simulated combustion experiment to measure the combustion characteristic parameters of the material. After the experiment is completed, professionals will verify the experimental data and generate a complete report on the material combustion characteristics. The verified combustion characteristic parameters are entered into the dynamic combustible material database, and relevant information such as update time, reviewer, and experimenter is recorded to form an update log.

Images