A super high-rise building hanging truss type fire extinguishing equipment

Abstract

The application relates to the technical field of fire extinguishing, and discloses a super high-rise building hanging truss type fire extinguishing equipment which comprises a fire extinguishing mechanism and a control module, the control module is electrically connected with the fire extinguishing mechanism, the control module is used for controlling the fire extinguishing mechanism, the fire extinguishing mechanism (1) comprises a hoisting component, a fire extinguishing component, a terahertz emitter, a temperature sensor, a smoke sensor, a wind speed sensor and a driving component, the hoisting component is fixedly connected with a building roof, the fire extinguishing component is fixedly connected with the hoisting component, the driving component is fixedly connected with the hoisting component and the fire extinguishing component respectively, and the terahertz emitter, the temperature sensor, the smoke sensor and the wind speed sensor are all located on the fire extinguishing component. The application improves the comprehensive sensing capability under a severe fire environment, reduces the uncertainty of fire source positioning, and improves the precision and efficiency of fire extinguishing operation.

Description

Technical Field

[0001] This invention relates to the field of fire extinguishing technology, and in particular to a suspended truss-type fire extinguishing device for ultra-high-rise buildings. Background Technology

[0002] With the acceleration of urbanization, super high-rise buildings have become an important part of the modern urban skyline. These buildings are characterized by their great height, complex structure, high population density, and diverse functional areas. Once a fire occurs, it can easily cause significant casualties and property damage. Therefore, fire fighting in super high-rise buildings has become a globally recognized technical challenge in the field of fire protection.

[0003] However, current mainstream firefighting methods for super high-rise buildings have many limitations and cannot meet the needs of efficient firefighting: First, the firefighting capacity of ground fire trucks is insufficient. The maximum operating height of conventional fire ladder trucks is generally between 50 and 80 meters, and even high-end ladder trucks cannot effectively exceed 100 meters, making it impossible to cover the middle and upper floors of super high-rise buildings. The spray height of fire monitors is greatly affected by water pressure and wind force. The pressure decreases and atomization is severe during the upward flow of water, resulting in insufficient firefighting impact and coverage when reaching the fire point. Second, helicopter firefighting has significant risks and limitations. The interaction between the high-temperature hot airflow at the fire scene and the airflow from the helicopter rotor can exacerbate the spread of the fire and cause the helicopter to become unstable. In addition, it is greatly limited by weather conditions, and its limited water carrying capacity makes it impossible to continuously extinguish fires. Third, drone firefighting technology is not yet mature. Due to limitations in materials and power systems, its carrying capacity is weak, making it unable to carry large-volume fire extinguishing media and high-power equipment. High-temperature environments can damage drone electronic components, batteries, and structural materials, leading to loss of control and crashes. Furthermore, airflow can disrupt flight stability, making it difficult to accurately locate the fire point. The operating altitude and firefighting efficiency cannot meet the requirements. Summary of the Invention

[0004] The purpose of this invention is to provide a suspended truss-type fire extinguishing device for ultra-high-rise buildings to solve the problems existing in the prior art.

[0005] To achieve the above objectives, the present invention provides the following solution: The present invention provides a suspended truss-type fire extinguishing device for ultra-high-rise buildings, comprising: A fire extinguishing mechanism (1) and a control module, wherein the control module is electrically connected to the fire extinguishing mechanism (1) and is used to control the fire extinguishing mechanism (1); the fire extinguishing mechanism (1) includes: The system includes a hoisting component, a fire extinguishing component, a terahertz transmitter, a temperature sensor, a smoke sensor, a wind speed sensor, and a drive component. The hoisting component is fixedly connected to the roof of the building, the fire extinguishing component is fixedly connected to the hoisting component, and the drive component is fixedly connected to both the hoisting component and the fire extinguishing component. The terahertz generator, the temperature sensor, the smoke sensor, and the wind speed sensor are all located on the fire extinguishing component. The control module includes: The sensing unit is configured to acquire terahertz image data, target distance data and ambient temperature data of the fire area based on the terahertz transmitter, and combine them with the real-time pose data of the fire extinguishing component to generate a sparse sensing dataset with spatiotemporal labels. The fire source inversion unit is configured to receive the sparse sensing dataset, construct a physical information neural network, use the fire source center coordinates and heat release rate as trainable variables of the network, and invert the fire source parameters by minimizing the total loss function that includes data fitting loss terms and physical equation residual loss terms. The water spray control unit is configured to receive the fire source parameters, establish a water flow trajectory model in the air, calculate the optimal spray angle of multiple water nozzles on the fire extinguishing truss based on the fire source parameters, generate a periodic oscillation pattern of each water nozzle, and convert the optimal spray angle and periodic oscillation pattern into control signal output. The early warning unit is configured to monitor the status of the fire source in real time during the water spraying process and determine whether the fire source has been extinguished.

[0006] Furthermore, the hoisting components include a track-mounted unmanned vehicle, a winch, and a telescopic boom. The track-mounted unmanned vehicle is fixedly connected to the roof of the building. The winch is mounted on the track-mounted unmanned vehicle and connected to the fire extinguishing components via the telescopic boom. The winch is used to drive the telescopic boom to move, and the telescopic boom is fixedly connected to the track-mounted unmanned vehicle.

[0007] Furthermore, the fire extinguishing component includes a truss body, several bidirectional swing nozzles, and a water supply pipeline. The truss body is fixedly connected to the winch via the telescopic boom. Several bidirectional swing nozzles are provided on the truss body. One end of the water supply pipeline is connected to a fire water tank, and the other end of the water supply pipeline is connected to the truss body.

[0008] Furthermore, when generating sparse-aware datasets with spatiotemporal labels, the following steps are included: The three-dimensional position coordinates, pitch angle, roll angle and yaw angle of the fire extinguishing truss are acquired in real time as the real-time pose data; Acquire terahertz reflection intensity images and obtain target distance values ​​by phase method; Ambient temperature data is collected by an array of temperature sensors integrated on the fire extinguishing truss; The terahertz image data, distance value, and ambient temperature data of each scanning point are spatiotemporally aligned with the truss pose data at the corresponding time. Based on the truss pose and the installation position and orientation of the terahertz transmitter, calculate the spatial coordinates of each scanning point in the building coordinate system; Generate a sparse sensing dataset containing timestamps, spatial coordinates, terahertz image features, distance values, and ambient temperature values.

[0009] Furthermore, when obtaining the fire source parameters through inversion, the following are included: A physical information neural network is constructed with spatial coordinates and time coordinates as inputs. The output of the physical information neural network includes temperature, pressure, velocity vector and smoke concentration. The fire dynamics physical equations, including the discretized forms of the Navier-Stokes equations, energy equations, and component transport equations, are embedded into the neural network as physical residual outputs. Construct a total loss function, which includes a data fitting loss term, a physical equation residual loss term, and a boundary condition loss term; The coordinates of the fire source center and the heat release rate are initialized as trainable variables, and gradient optimization is performed simultaneously with the weights of the physical information neural network to minimize the total loss function. Output the fire source parameters obtained by inversion, including the three-dimensional coordinates of the fire source center, the heat release rate, and the equivalent radius of the fire source.

[0010] Furthermore, when minimizing the total loss function, which includes both the data fitting loss term and the physical equation residual loss term, the following is included: The data fitting loss term is constructed, which includes at least a terahertz image loss term to constrain the temperature field output by the physical information neural network to be consistent with the high temperature region identified by the terahertz image at the measurement point; a distance loss term to constrain the fire source location predicted by the physical information neural network to be consistent with the distance value obtained by terahertz ranging; and an ambient temperature loss term to constrain the ambient temperature output by the physical information neural network to be consistent with the measured ambient temperature value. Construct the residual loss term of the physical equation, which is the sum of the residuals calculated after substituting the output of the physical information neural network into each physical equation; The boundary condition loss term is constructed as a constraint applied based on the positions of walls, doors, and windows in the building information model.

[0011] Furthermore, when receiving the fire source parameters and establishing a model of the water flow trajectory in the air, the process includes: Obtain water supply pressure parameters and determine water outlet velocity; Real-time wind speed and direction data are obtained by wind speed and direction sensors integrated on the fire extinguishing truss. Based on the water flow outlet velocity, wind speed and direction, and gravitational acceleration, a differential equation or analytical expression for the water flow trajectory is established as the motion trajectory model.

[0012] Furthermore, when calculating the optimal spray angle of multiple nozzles on the fire extinguishing truss based on the fire source parameters, the following steps are included: Based on the three-dimensional coordinates of the fire source center and the equivalent radius of the fire source output by the fire source inversion module, and combined with the motion trajectory model, the horizontal swing angle and pitch angle of multiple bidirectional swing nozzles on the fire extinguishing truss are respectively solved, wherein the multiple bidirectional swing nozzles are located at both ends and the middle of the fire extinguishing truss. The swing amplitude of each nozzle is determined based on the equivalent radius of the fire source, and the swing frequency of each nozzle is determined based on the fire intensity. Control parameters are generated to make each nozzle swing periodically in the horizontal and vertical directions. The flow rate of the three water nozzles is dynamically allocated according to the fire source parameters, with the two end nozzles focusing on covering both sides of the fire source and the middle nozzle focusing on covering the center of the fire source.

[0013] Furthermore, the water spray control unit is also configured to: during the water spraying process, acquire terahertz images and distance data of the fire source area in real time through the terahertz sensing and positioning module, and monitor the fire source shape and temperature changes; When the rate of temperature drop in the fire source area is detected to be lower than the preset threshold or a temperature rebound occurs, the injection parameters are recalculated or the swing strategy is adjusted.

[0014] Furthermore, determining whether a fire source has been extinguished includes: Real-time acquisition of terahertz image sequences and distance data sequences of the fire source area; Extract the area, highest temperature value, and temperature distribution characteristics of the high-temperature region from the terahertz image sequence; When the area of ​​the high-temperature region is continuously smaller than the first preset threshold, the highest temperature value is continuously lower than the second preset threshold, and the temperature distribution shows a uniform downward trend, it is determined that the fire source has been extinguished.

[0015] This invention discloses the following technical effects: By leveraging the dual physical properties of terahertz waves—the ability to penetrate smoke and sensitivity to temperature—it achieves "visual penetration" in extreme fire environments where visible light is completely ineffective and infrared thermography is severely interfered with by smoke particle scattering. This enhances the ability to capture the true shape and outline of the fire source, eliminating the limitations imposed by harsh conditions such as dense smoke, darkness, and high-temperature water mist, thus improving the environmental adaptability and reliability of sensing equipment in complex fire scenes. Simultaneously, terahertz imaging technology can clearly distinguish high-temperature fire sources from the background environment, avoiding the risks of misjudgment and missed judgment caused by data quality degradation in traditional multi-sensor fusion under harsh conditions. This provides a high-quality, high-confidence raw data foundation for subsequent location and firefighting decisions. Secondly, traditional location methods often face significant location errors and uncertainties when fire data is sparse and observation points are limited. This invention deeply integrates sparse image features, distance information, and ambient temperature data obtained from terahertz scanning with physical equations describing fire dynamics. By using the fire source center coordinates and heat release rate as trainable variables in a neural network, and leveraging the residuals of physical equations as strong constraints to guide network training, the goal of retrieving the precise three-dimensional coordinates and heat release rate of the fire source from only a small number of sparse measurement points was achieved. This reduces the reliance on deploying a large number of dense sensors within the fire scene, simplifies the system structure, and lowers implementation costs. Simultaneously, the embedding of physical equations suppresses sensor noise and environmental interference, enhancing the robustness and physical consistency of the inversion results, ensuring accurate positioning even under sparse data conditions. Furthermore, based on the inverted fire source parameters, intelligent adaptive control of the fire extinguishing actuators was achieved. According to the fire source center coordinates, the equivalent radius of the fire source, and real-time wind speed and direction data, the trajectory of the water flow in the air can be calculated. For the three bidirectional oscillating nozzles located at both ends and the middle of the fire extinguishing truss, the optimal horizontal oscillation angle and pitch angle are solved, improving the accuracy of the spray and avoiding blind sweeping. Meanwhile, by generating periodic oscillation patterns for each nozzle and dynamically adjusting the oscillation amplitude and frequency according to the size of the fire source, three-dimensional cross-coverage of the fire source area is achieved, improving the fire source coverage efficiency and the overall effectiveness of fire extinguishing operations. Attached Figure Description

[0016] The accompanying drawings, which form part of this application, are used to provide a further understanding of this application. The illustrative embodiments and descriptions of this application are used to explain this application and do not constitute an undue limitation of this application. In the drawings: Figure 1 A schematic diagram of the overall structure of the suspended truss-type fire extinguishing equipment for super high-rise buildings provided in an embodiment of the present invention; Figure 2 A functional block diagram of a suspended truss-type fire extinguishing device for super high-rise buildings provided in an embodiment of the present invention.

[0017] In the diagram: 1. Fire extinguishing mechanism; 110. Lifting component; 1101. Rail-mounted unmanned vehicle; 1102. Winch; 1103. Telescopic boom; 120. Fire extinguishing component; 1201. Truss main body; 1202. Bidirectional swing water nozzle; 1203. Water supply pipeline; 130. Terahertz transmitter; 140. Temperature sensor; 150. Smoke sensor; 160. Wind speed sensor; 170. Drive component. Detailed Implementation

[0018] The technical solutions of the embodiments of the present invention will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of the present invention, and not all embodiments. Based on the embodiments of the present invention, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of the present invention.

[0019] To make the above-mentioned objects, features and advantages of the present invention more apparent and understandable, the present invention will be further described in detail below with reference to the accompanying drawings and specific embodiments.

[0020] Exemplary embodiments of the present disclosure will now be described in more detail with reference to the accompanying drawings. While exemplary embodiments of the present disclosure are shown in the drawings, it should be understood that the present disclosure may be implemented in various forms and should not be limited to the embodiments set forth herein. Rather, these embodiments are provided to enable a more thorough understanding of the present disclosure and to fully convey the scope of the disclosure to those skilled in the art. It should be noted that, unless otherwise specified, embodiments and features in the embodiments of the present invention can be combined with each other. The present invention will now be described in detail with reference to the accompanying drawings and embodiments.

[0021] In some embodiments of this application, see Figure 1 , Figure 2 As shown, a suspended truss-type fire extinguishing device for ultra-high-rise buildings includes: The fire extinguishing mechanism (1) and the control module are electrically connected to the fire extinguishing mechanism (1) and the control module is used to control the fire extinguishing mechanism (1); the fire extinguishing mechanism (1) includes: The system includes a hoisting component, a fire extinguishing component, a terahertz transmitter, a temperature sensor, a smoke sensor, a wind speed sensor, and a drive component. The hoisting component is fixedly connected to the roof of the building, the fire extinguishing component is fixedly connected to the hoisting component, and the drive component is fixedly connected to both the hoisting component and the fire extinguishing component. The terahertz generator, temperature sensor, smoke sensor, and wind speed sensor are all located on the fire extinguishing component. The control module includes: The sensing unit is configured to acquire terahertz image data, target distance data and ambient temperature data of the fire area based on the terahertz transmitter, and combine them with the real-time pose data of the fire extinguishing components to generate a sparse sensing dataset with spatiotemporal labels. The fire source inversion unit is configured to receive a sparse sensing dataset, construct a physical information neural network, use the fire source center coordinates and heat release rate as trainable variables of the network, and obtain the fire source parameters by minimizing the total loss function that includes data fitting loss terms and physical equation residual loss terms. The water spray control unit is configured to receive fire source parameters, establish a water flow trajectory model in the air, calculate the optimal spray angle of multiple water nozzles on the fire extinguishing truss based on the fire source parameters, generate the periodic oscillation pattern of each water nozzle, and convert the optimal spray angle and periodic oscillation pattern into control signal output. The early warning unit is configured to monitor the status of the fire source in real time during the water spraying process and determine whether the fire source has been extinguished.

[0022] Specifically, the fire extinguishing system is installed on the roof of the building. When a fire occurs, after determining the specific floor, the fire extinguishing unit is lowered to the corresponding floor position by the hoisting component. The sensing unit acquires data of the fire area, determines the fire source parameters based on the data of the fire area, and then generates a water flow trajectory model based on the fire source parameters to determine the spray angle of the water nozzles on the fire extinguishing unit and then carry out the fire extinguishing operation. At the same time, the early warning unit monitors the fire source status in real time to determine whether the fire source has been extinguished.

[0023] It is understandable that terahertz waves, with their dual physical properties of penetrating smoke and being sensitive to temperature, can achieve "visual penetration" in extreme fire environments where visible light is completely ineffective and infrared thermography is severely interfered with by smoke particles. This enhances the ability to capture the true shape and outline of the fire source, freeing it from the interference of harsh conditions such as dense smoke, darkness, and high-temperature water mist, thus improving the environmental adaptability and reliability of sensing equipment in complex fire scenes. Simultaneously, terahertz imaging technology can clearly distinguish high-temperature fire sources from the background environment, avoiding the risks of misjudgment and missed judgment caused by data quality degradation in traditional multi-sensor fusion under harsh conditions, providing a high-quality, high-confidence raw data foundation for subsequent location and firefighting decisions. Secondly, traditional location methods often face significant location errors and uncertainties when fire data is sparse and observation points are limited. This invention deeply integrates sparse image features, distance information, and ambient temperature data obtained from terahertz scanning with physical equations describing fire dynamics. By using the fire source center coordinates and heat release rate as trainable variables in a neural network, and leveraging the residuals of physical equations as strong constraints to guide network training, the goal of retrieving the precise three-dimensional coordinates and heat release rate of the fire source from only a small number of sparse measurement points was achieved. This reduces the reliance on deploying a large number of dense sensors within the fire scene, simplifies the system structure, and lowers implementation costs. Simultaneously, the embedding of physical equations suppresses sensor noise and environmental interference, enhancing the robustness and physical consistency of the inversion results, ensuring accurate positioning even under sparse data conditions. Furthermore, based on the inverted fire source parameters, intelligent adaptive control of the fire extinguishing actuators was achieved. According to the fire source center coordinates, the equivalent radius of the fire source, and real-time wind speed and direction data, the trajectory of the water flow in the air can be calculated. For the three bidirectional oscillating nozzles located at both ends and the middle of the fire extinguishing truss, the optimal horizontal oscillation angle and pitch angle are solved, improving the accuracy of the spray and avoiding blind sweeping. Meanwhile, by generating periodic oscillation patterns for each nozzle and dynamically adjusting the oscillation amplitude and frequency according to the size of the fire source, three-dimensional cross-coverage of the fire source area is achieved, improving the fire source coverage efficiency and the overall effectiveness of fire extinguishing operations.

[0024] In some embodiments of this application, the hoisting components include a track-mounted unmanned vehicle, a winch, and a telescopic boom. The track-mounted unmanned vehicle is fixedly connected to the roof of a building. The winch is mounted on the track-mounted unmanned vehicle and connected to the fire extinguishing components via the telescopic boom. The winch is used to drive the telescopic boom to move. The telescopic boom is fixedly connected to the track-mounted unmanned vehicle.

[0025] In some embodiments of this application, the fire extinguishing component includes a truss body, several bidirectional swing nozzles, and a water supply pipeline. The truss body is fixedly connected to a winch via a telescopic boom. Several bidirectional swing nozzles are provided on the truss body. One end of the water supply pipeline is connected to a fire water tank, and the other end of the water supply pipeline is connected to the truss body.

[0026] Specifically, the track-mounted unmanned vehicle adopts a modular design, equipped with a high-precision drive motor and guiding mechanism. It can travel smoothly along a pre-set circular or linear track on the roof. The track is made of high-strength steel and is firmly connected to the main roof structure using expansion bolts. The unmanned vehicle integrates a diesel generator set with an automatic start device, which can be remotely started upon receiving a fire alarm signal to provide stable power support for the winch, boom drive mechanism, and remote monitoring equipment. The winch adopts a variable frequency speed control design and has a reserved dedicated water intake interface for the rooftop fire water tank. It is equipped with an electric control valve that can remotely control the water intake switch. The power supply line is integrated into the fire-fighting special steel wire rope bundle and is lowered synchronously with the steel wire rope. The fire extinguishing component is the core execution mechanism for fire fighting. The main body adopts a truss structure design, which is lightweight, high-strength, and high-temperature resistant. It includes the truss main body, three bidirectional swing water nozzles (one at each end of the truss and one in the middle), branch pipelines, and multiple sensors. The device uses the truss itself as the main water supply pipe, eliminating the need for a traditional independent fire-fighting main water pipe design. During firefighting operations, the water circulates within the truss components, achieving active cooling. It adopts a modular splicing design with a triangular cross-section, providing excellent bending and torsional resistance. All three nozzles are made of stainless steel and equipped with a high-precision servo drive mechanism and small hydraulic push rods (controlling horizontal and vertical movement respectively), allowing for independent and flexible horizontal and vertical swinging. It features a high-pressure atomization design, which can independently adjust the columnar water flow (suitable for long-distance firefighting and penetrating fire) and atomized water flow (suitable for large-area cooling and suppressing fire spread) according to the fire type.

[0027] The bidirectional swing sprinklers symmetrically installed at both ends of the truss can adjust the difference in water flow and velocity on both sides through commands output by the control module to create a reverse thrust difference, thereby achieving horizontal stability adjustment and swing control of the truss. The branch pipelines are welded and fixed with high-temperature resistant joints compatible with the truss material, and the connection parts adopt a high-pressure sealing design. High-precision electric flow control valves are equipped on both the main pipeline and the branch pipelines. The flow and velocity of each sprinkler can be independently adjusted through the control computer or remote control system. Fire early warning response and remote system start-up are also available. After the building's internal fire alarm system is triggered, the remote monitoring equipment of the roof hoisting system is automatically activated. After the command center confirms the fire situation and the precise floor of the fire through the remote monitoring platform, it sends a start command. The roof generator automatically starts to supply power, the electric control valve of the fire water tank water intake interface opens, the high-pressure booster pump starts preheating, the boom slowly unfolds, the winch enters the standby state, and then the fire extinguishing system is lowered to accurately reach the fire floor. After the boom is deployed, the winch starts lowering the fire extinguishing system, with the power supply line lowered synchronously along with the special fire-fighting steel wire rope. The winch maintains a stable lifting speed. Based on wind speed sensor data, the two water nozzles are activated and the flow rate is adjusted to create a balanced thrust and maintain horizontal stability. When lowered to the vicinity of the fire floor, the height is adjusted, and the water nozzles are aimed at the fire area and surrounding key areas. Water supply is activated to precisely suppress the fire. After the fire extinguishing system is positioned, the three water nozzles are remotely activated, and the high-pressure booster pump enters its rated operating state. The water flow pattern, swing angle, and flow rate of each nozzle are independently adjusted according to the fire situation.

[0028] Understandably, the coordinated use of rail-mounted unmanned vehicles, winches, and telescopic booms enhances the overall deployment capability of firefighting equipment. The rail-mounted unmanned vehicles, with their modular design and high-precision drive mechanism, can travel smoothly along pre-set circular or straight tracks on the rooftop, improving the equipment's mobility and positioning accuracy on the rooftop and allowing firefighting components to be quickly positioned at the optimal working point above the burning floor. The winch, employing a variable frequency speed control design, combined with the telescopic boom, enables precise and stable vertical lowering of firefighting components, reducing the risk of swaying during hoisting and improving the safety and controllability of high-altitude operations. The diesel generator set integrated into the unmanned vehicle is equipped with an automatic start device, enabling remote start upon receiving a fire alarm signal, providing stable and independent power support for the entire system. This eliminates reliance on the building's own power supply system, improving the system's emergency response capability and reliability. The truss main body utilizes lightweight, high-strength, and high-temperature-resistant materials with a modular splicing design. Its triangular cross-section gives the truss excellent bending and torsional resistance, enhancing its structural stability under wind loads and jet reaction forces. The truss itself directly serves as the main water supply pipeline, eliminating the traditional independent fire-fighting main water pipe design. This not only simplifies the pipeline layout and reduces overall weight but also enables water circulation within the truss components, actively cooling the truss itself during firefighting operations and improving the equipment's continuous operation and service life in high-temperature fire environments. Branch pipelines are welded and fixed with high-temperature resistant joints and equipped with a high-pressure sealing design, ensuring the reliability and sealing of the water supply system under high pressure. Three bidirectional swing nozzles (one at each end of the truss and one in the middle) are all made of stainless steel and equipped with a high-precision servo drive mechanism and a small hydraulic push rod, enabling independent and flexible swinging in both horizontal and vertical directions, improving the flexibility and accuracy of the spray coverage. The symmetrically installed water nozzles at both ends of the truss can adjust the flow rate and velocity difference of the water on both sides through control commands, thereby creating a reverse thrust difference. This achieves horizontal stability adjustment and swing control of the truss in the water, eliminating the need for additional attitude adjustment mechanisms and improving the equipment's dynamic balance and anti-interference capabilities in the air. The water nozzles adopt a high-pressure atomization design, which can independently adjust the columnar water flow (suitable for long-distance fire extinguishing and penetrating fire) and the atomized water flow (suitable for large-area cooling and suppressing fire spread) according to the fire type, enhancing adaptability to different fire scenarios.

[0029] In some embodiments of this application, generating a sparse-aware dataset with spatiotemporal labels includes: The three-dimensional position coordinates, pitch angle, roll angle and yaw angle of the fire extinguishing truss are acquired in real time as real-time pose data; Acquire terahertz reflection intensity images and obtain target distance values ​​by phase method; Ambient temperature data is collected by an array of temperature sensors integrated into the fire extinguishing truss; The terahertz image data, distance value, and ambient temperature data of each scanning point are spatiotemporally aligned with the truss pose data at the corresponding time. Based on the truss pose and the installation position and orientation of the terahertz transmitter, calculate the spatial coordinates of each scanning point in the building coordinate system; Generate a sparse sensing dataset containing timestamps, spatial coordinates, terahertz image features, distance values, and ambient temperature values.

[0030] Specifically, at each measurement time t, data fusion using inertial measurement units integrated on the fire extinguishing truss is used to obtain the three-dimensional position coordinates of the truss coordinate system origin in the world coordinate system (i.e., the building coordinate system) and the truss's attitude angles. Definition: The position vector of the origin of the truss coordinate system in the world coordinate system; attitude angles are represented by Euler angles: yaw angle. Pitch angle Roll angle Define the rotation matrix from the truss coordinate system to the world coordinate system according to the ZYX rotation order. : ,in, , , . Satisfy vectors in any truss coordinate system Its representation in the world coordinate system is At time t, the terahertz transmitter and receiver are controlled to scan the target area to acquire a frame of terahertz reflection intensity image. Simultaneously, the distance value d(t) of the high-temperature target point corresponding to this frame image is obtained by measuring the phase difference between the emitted and reflected terahertz waves using the phase method. To calculate the distance: The ambient temperature value is collected at time t by an array of temperature sensors integrated on the fire extinguishing truss. The temperature sensor array can be arranged at different positions on the truss, and the average or representative value can be taken as the ambient temperature at that moment.

[0031] To spatiotemporally align the sensor data at each measurement point with the truss pose, it is necessary to record the precise time *t* corresponding to each measurement point and ensure that the timestamps of all data are synchronized. Then, based on the known installation parameters of the terahertz sensor on the truss, the spatial coordinates of the target point in the world coordinate system are calculated. Let the position offset vector of the origin of the terahertz sensor coordinate system relative to the origin of the truss coordinate system be... (Constant value, obtained from installation location calibration), the rotation matrix from the sensor coordinate system to the truss coordinate system is: (Constant value, obtained from installation orientation calibration). Assuming the sensor emits a beam along the Z-axis of its own coordinate system, the sensor's position in the world coordinate system is... and the unit vector of the launch direction They are respectively: as well as Then the coordinates of the target point (i.e., the terahertz wave reflection point) in the world coordinate system are... for: Next, all the above information is combined to generate a data record for each measurement point. The final sparse sensing dataset is then obtained. : ,in, Let be the position vector of the origin of the truss coordinate system in the world coordinate system at time t. Let c be the rotation matrix from the truss coordinate system to the world coordinate system at time t, and c be the speed of light, taking values ​​of 10 ... meters per second (m / s) is a physical constant. The frequency of the terahertz wave.

[0032] Understandably, terahertz waves possess the dual physical properties of penetrating smoke and being sensitive to temperature. This allows for the stable acquisition of terahertz reflection intensity images reflecting the true shape of the fire source in fire environments where visible light is completely ineffective and infrared thermography is severely interfered with by smoke particle scattering. Simultaneously, the phase method is used for distance measurement, calculating the target distance by measuring the phase difference between the emitted and reflected waves. Compared to the traditional pulse-time-of-flight method, the phase method achieves higher ranging resolution and accuracy under the same hardware conditions, improving the quality of information acquired regarding the fire source's location and distance. The strict mathematical relationship between phase difference and distance ensures the physical accuracy and repeatability of the distance measurement. Furthermore, by simultaneously recording the real-time pose data of the fire extinguishing truss, terahertz image data, distance measurements, and ambient temperature data, all observation information corresponds to the same physical moment, eliminating time misalignment caused by differences in sensor sampling frequencies or communication delays. This reduces uncertainty in the data fusion process and provides a reliable time reference for subsequent calculations of the target point's spatial coordinates.

[0033] In some embodiments of this application, the inversion to obtain the fire source parameters includes: A physical information neural network is constructed with spatial and temporal coordinates as inputs. The output of the physical information neural network includes temperature, pressure, velocity vector, and smoke concentration. Fire dynamics physical equations are embedded into neural networks. These equations include discretized forms of the Navier-Stokes equations, energy equations, and component transport equations, which are then output as physical residuals. Construct a total loss function, which includes a data fitting loss term, a physical equation residual loss term, and a boundary condition loss term; The coordinates of the fire source center and the heat release rate are initialized as trainable variables, and gradient optimization is performed simultaneously with the weights of the physical information neural network to minimize the total loss function. The output inverted fire source parameters include the three-dimensional coordinates of the fire source center, the heat release rate, and the equivalent radius of the fire source.

[0034] In some embodiments of this application, minimizing the total loss function, which includes a data fitting loss term and a physical equation residual loss term, includes: Construct a data fitting loss term, which includes at least a terahertz image loss term to constrain the temperature field output by the physical information neural network to be consistent with the high temperature region identified by the terahertz image at the measurement point; a distance loss term to constrain the fire source location predicted by the physical information neural network to be consistent with the distance value obtained by terahertz ranging; and an ambient temperature loss term to constrain the ambient temperature output by the physical information neural network to be consistent with the measured ambient temperature value. Construct a residual loss term for the physical equations, which is the sum of the residuals calculated after substituting the output of the physical information neural network into each physical equation; Construct boundary condition loss terms, which are constraints applied based on the locations of walls, doors, and windows in the building information model.

[0035] Specifically, a neural network is constructed, with spatial coordinates as input. Given time t, the output is the physical quantities of the fire scene: ,in, For temperature field, For pressure field, These are the three components of the velocity field. The smoke concentration, This is the set of weight parameters for the neural network. The neural network can calculate the partial derivatives of each output with respect to the input through automatic differentiation, providing a foundation for subsequent residual calculations of the physical equations.

[0036] Then, the partial differential equations describing fire dynamics are used as physical constraints, and internal sampling points are then used within the computational domain. The following residual terms are defined above. The Boussinesq approximation is used, assuming that density is constant outside the buoyancy term of the momentum equation. The continuity equation simplifies to an incompressible form, the continuity equation (mass conservation): Momentum equation (Navier-Stokes equations, considering buoyancy): X direction: ; Y direction: ; Z direction: .in, Kinematic viscosity, gravitational acceleration vector The coefficient of thermal expansion is This is a reference temperature.

[0037] Energy equation (including heat release from the ignition source): ,in, Where is the thermal diffusivity, and k is the thermal conductivity. For isobaric specific heat capacity, Heat release rate from the heat source Coordinates of the fire source center To smooth the Dirac function, used to approximate a point source, the component transport equation (smoke concentration) is as follows: Where D is the smoke mass diffusion coefficient. This is the smoke production rate coefficient. The physical residual loss term is defined as the sum of the squares of the residuals of all equations at all internal sampling points: Then, a data fitting loss term is constructed: based on the sparse sensing dataset. Define the following loss term: Terahertz image loss term: Temperature at the measurement point estimated using terahertz images. Constrain the network to output consistent temperatures: ,in, This is the temperature output of the network at the measurement point.

[0038] Distance loss term: measurement point Located on the surface of a fire source or in a high-temperature area, its distance from the center of the fire source should be equal to the radius of the fire source. ,in, For the Euclidean norm, As the center of the fire, The equivalent radius of the fire source.

[0039] Ambient temperature loss item: Truss location The network output temperature should be consistent with the measured ambient temperature. .

[0040] The data fitting loss term is the weighted sum of the above three terms: ,in, These are the weighting coefficients. Next, a boundary condition loss term is constructed, based on structural information such as walls, doors, and windows provided by Building Information Modeling (BIM), at the boundary points. Physical constraints are applied. The normal velocity at the solid wall is zero, the pressure at the opening is atmospheric pressure, etc. The boundary condition loss term is the sum of the squares of the constraint residuals at each boundary point: ,in, For the boundary condition residuals at the j-th boundary point (such as the normal velocity at the solid wall), then construct the total loss function. ,in, and These are the weighting coefficients.

[0041] Next, the fire source parameters are initialized as trainable variables, including the fire source center coordinates, heat release rate, and equivalent fire source radius (e.g., as additional inputs to the neural network or directly as optimization variables). These variables, along with the network weights, participate in subsequent optimization. An automatic differentiation and gradient descent optimization algorithm is used to iteratively update the neural network weights and fire source parameters, minimizing the total loss until convergence. Finally, after optimization, the final fire source center coordinates, heat release rate, and equivalent fire source radius are output as inversion results for subsequent water spray control. For the total loss, This represents the boundary condition loss, where n is the boundary normal vector. These are the coordinates of the boundary points. Number of boundary sampling points The ambient temperature at the i-th measurement point Terahertz imaging for temperature estimation Let i be the timestamp of the i-th observation point. Let N be the coordinates of the i-th observation point, and N be the number of measurement points. For physical residual loss, For the residuals of each equation, For physical residual sampling points, This represents the number of physical residual sampling points.

[0042] Understandably, this invention constructs a neural network with spatial and temporal coordinates as input, directly outputting complete fire field physical fields such as temperature, pressure, velocity vectors, and smoke concentration. Through automatic differentiation, the neural network can accurately calculate the partial derivatives of each output with respect to the input, providing a mathematical foundation for subsequently embedding complex fire dynamics equations. This allows the network to deduce a globally consistent physical field distribution from a limited number of observations, even with a limited number of measurement points and sparse data, constrained by the physical equations. This improves the efficiency of utilizing sparse sensing datasets and reduces dependence on dense sensor arrays. The fire source center coordinates are directly incorporated as trainable variables into the optimization process. Through the constraint of the distance loss term, the distance between each measurement point and the fire source center is forced to be equal to the equivalent radius of the fire source. This geometric constraint fully utilizes the high-precision distance information obtained from terahertz ranging, establishing a direct mathematical connection between sparse measurement points and the fire source location, improving the inversion accuracy of the fire source center coordinates and reducing the positioning ambiguity caused by sparse observation data. Thirdly, regarding improving the reliability of the fire source heat release rate inversion, the scheme embeds the heat release rate into the neural network through the fire source term in the energy equation. This allows the heat release rate to be used as a trainable variable, directly participating in the residual calculation of the physical equations. The optimization of its value is constrained by the overall temperature field distribution and the law of energy conservation. Compared to traditional methods that rely solely on data fitting, this physical embedding mechanism improves the reliability of fire source intensity inversion and effectively reduces the estimation bias of the heat release rate caused by data noise or observation blind spots. Fourth, a comprehensive physical equation residual loss term is constructed. By using the discretized forms of the continuity equation, momentum equation, energy equation, and component transport equation as constraints, the output of the neural network is forced to satisfy the laws of mass conservation, momentum conservation, energy conservation, and mass transport at its internal sampling points. This mechanism significantly improves the intrinsic consistency between the inverted temperature field, velocity field, and smoke concentration field, making the entire fire scene description physically self-consistent and effectively reducing the non-physical predictions that may arise from purely data-driven methods.

[0043] In some embodiments of this application, when receiving fire source parameters and establishing a water flow trajectory model in the air, the following steps are included: Obtain water supply pressure parameters and determine water outlet velocity; Real-time wind speed and direction data are obtained by integrating wind speed and direction sensors onto the fire extinguishing truss. Based on the water outlet velocity, wind speed and direction, and gravitational acceleration, establish the differential equation or analytical expression of the water flow trajectory as a motion trajectory model.

[0044] In some embodiments of this application, calculating the optimal spray angle of multiple nozzles on the fire extinguishing truss based on fire source parameters includes: Based on the three-dimensional coordinates of the fire source center and the equivalent radius of the fire source output by the fire source inversion module, and combined with the motion trajectory model, the horizontal swing angle and pitch angle of multiple bidirectional swing nozzles on the fire extinguishing truss are solved respectively. The multiple bidirectional swing nozzles are located at both ends and the middle of the fire extinguishing truss. The swing amplitude of each nozzle is determined based on the equivalent radius of the fire source, and the swing frequency of each nozzle is determined based on the fire intensity. Control parameters are generated to make each nozzle swing periodically in the horizontal and vertical directions. The flow rate of the three water nozzles is dynamically allocated according to the fire source parameters, with the two end nozzles focusing on covering both sides of the fire source and the middle nozzle focusing on covering the center of the fire source.

[0045] Specifically, sensors acquire water outlet velocity and real-time wind speed and direction data, and then a water flow trajectory model is established for the nozzle. Its export location is The export speed is Horizontal swing angle (Rotating counterclockwise from the positive x-axis of the world coordinate system), pitch angle (Upward is positive), the position of the water droplet at time t satisfy: This is the water flow trajectory model, in which... The wind speed is calculated in real time. Then, the coordinates of the fire source center and the equivalent radius of the fire source are obtained. Based on the fire source parameters, the target landing points of the three water nozzles are set. Known exit location and target landing point Record the horizontal displacement: The horizontal angle needs to be solved. Pitch angle and flight time Satisfying the water flow trajectory model in When the target point is reached: The solution is obtained using numerical methods. In practical applications, wind acceleration is initially neglected. Obtain initial analytical values, then substitute them into the complete equation for iterative correction. Analytical solution without wind. The horizontal distance is obtained from the first two equations. And there are Substituting, we get: ,use and , can obtain information about The quadratic equation is solved to obtain the following result. And thus This analytical solution is used as the initial value for iteration, and finally the value of each nozzle is obtained. The swing amplitude of each nozzle is determined based on the equivalent radius of the fire source. Let the horizontal swing amplitude be... and pitch swing amplitude Its swing angle is: ,in, and This is a constant, calibrated based on nozzle characteristics. Then, it is determined based on fire intensity (in terms of heat release rate). Characterization), determining the oscillation frequency : ,in, The frequency coefficient is set to an upper limit based on mechanical frequency limiting. Using a sine wave as the oscillation mode, the real-time angle of the nozzle is: , ,in and This is the initial phase.

[0046] Understandably, the water outlet velocity is calculated by acquiring water supply pressure parameters, and a water flow trajectory model is established by combining real-time wind speed and direction data. The motion of water droplets under gravity, initial velocity, and wind field is described in the form of differential equations or analytical expressions, fully considering the interference of wind on the horizontal motion of water droplets. This improves the accuracy of predicting the water droplet landing point, enabling water spray control to be calculated based on the real physical environment and reducing spray deviations caused by neglecting wind loads. Secondly, based on the three-dimensional coordinates of the fire source center and the equivalent radius of the fire source output by the fire source inversion module, the horizontal swing angle and pitch angle of the three bidirectional swing nozzles on the fire extinguishing truss are solved using the motion trajectory model. For each nozzle, a system of nonlinear equations with the outlet position and target landing point as boundary conditions is established, and numerical methods are used to solve for the horizontal angle, pitch angle, and flight time. Specifically, by first ignoring wind acceleration to obtain analytical initial values, and then substituting them into the complete equations for iterative correction, this strategy of initial value plus iteration improves the solution efficiency and convergence stability of complex nonlinear equation systems, while reducing computational time and divergence risk. Third, the oscillation amplitude of each nozzle is determined based on the equivalent radius of the fire source. By introducing an oscillation amplitude proportional to the size of the fire source, the dynamic coverage range of the nozzles can adapt to the geometric scale of the fire source: when the fire source is large, the oscillation amplitude automatically increases to expand the coverage area; when the fire source is small, the oscillation amplitude decreases to concentrate the spray. This improves the adaptability to fire sources of different sizes and reduces the problem of insufficient or excessive coverage caused by changes in fire source size. Simultaneously, the oscillation frequency is determined based on the fire intensity, so that the more intense the fire, the higher the oscillation frequency, thereby accelerating the scanning and cooling speed of the fire source area and significantly improving the response capability to dynamic fires.

[0047] In some embodiments of this application, the water spray control unit is further configured to: during the water spraying process, acquire terahertz images and distance data of the fire source area in real time through the terahertz sensing and positioning module, and monitor the shape and temperature changes of the fire source; When the rate of temperature drop in the fire source area is detected to be lower than the preset threshold or a temperature rebound occurs, the injection parameters are recalculated or the swing strategy is adjusted.

[0048] In some embodiments of this application, determining whether a fire source has been extinguished includes: Real-time acquisition of terahertz image sequences and distance data sequences of the fire source area; Extract the area, maximum temperature value, and temperature distribution features of high-temperature regions from terahertz image sequences; When the area of ​​the high-temperature zone is continuously smaller than the first preset threshold, the highest temperature value is continuously lower than the second preset threshold, and the temperature distribution shows a uniform downward trend, it is determined that the fire source has been extinguished.

[0049] Understandably, during water spraying, the terahertz sensing and positioning module continuously acquires terahertz images and distance data of the fire source area in real time, enabling uninterrupted monitoring of the fire source's morphology and temperature changes. This mechanism transforms water spraying control from a one-way open-loop operation into a closed-loop system of "spraying-monitoring-adjustment." When the rate of temperature decrease in the fire source area is detected to be below a preset threshold or a temperature rebound occurs, the spraying parameters are recalculated or the oscillation strategy is adjusted. This enhances the ability to respond to sudden changes in fire intensity (such as fire reignition or accelerated spread) and reduces the risk of decreased fire extinguishing efficiency due to unchanged strategies caused by changes in fire intensity. Secondly, this invention not only focuses on a single temperature value but also extracts multi-dimensional information such as the area of ​​the high-temperature region, the highest temperature value, and temperature distribution characteristics through terahertz image sequences. The area of ​​the high-temperature region reflects the spatial spread range of the fire, the highest temperature value characterizes the intensity of the fire source core, and the temperature distribution characteristics reveal the uniformity and development direction of the fire. The joint monitoring of multi-dimensional features enhances the ability to comprehensively perceive the fire source status and avoids misjudgments that may result from relying solely on a single temperature indicator.

[0050] In summary, the beneficial effects of this invention are as follows: By leveraging the dual physical properties of terahertz waves—the ability to penetrate smoke and sensitivity to temperature—it achieves "visual penetration" in extreme fire environments where visible light is completely ineffective and infrared thermography is severely interfered with by smoke particle scattering. This enhances the ability to capture the true shape and outline of the fire source, eliminating the limitations imposed by harsh conditions such as dense smoke, darkness, and high-temperature water mist, thus improving the environmental adaptability and reliability of sensing equipment in complex fire scenes. Simultaneously, terahertz imaging technology can clearly distinguish between high-temperature fire sources and the background environment, avoiding the risks of misjudgment and missed judgment caused by data quality degradation in traditional multi-sensor fusion under harsh conditions. This provides a high-quality, high-confidence raw data foundation for subsequent location and firefighting decisions. Secondly, traditional location methods often face significant location errors and uncertainties when fire data is sparse and observation points are limited. This invention deeply integrates sparse image features, distance information, and ambient temperature data obtained from terahertz scanning with physical equations describing fire dynamics. By using the fire source center coordinates and heat release rate as trainable variables in a neural network, and leveraging the residuals of physical equations as strong constraints to guide network training, the goal of retrieving the precise three-dimensional coordinates and heat release rate of the fire source from only a small number of sparse measurement points was achieved. This reduces the reliance on deploying a large number of dense sensors within the fire scene, simplifies the system structure, and lowers implementation costs. Simultaneously, the embedding of physical equations suppresses sensor noise and environmental interference, enhancing the robustness and physical consistency of the inversion results, ensuring accurate positioning even under sparse data conditions. Furthermore, based on the inverted fire source parameters, intelligent adaptive control of the fire extinguishing actuators was achieved. According to the fire source center coordinates, the equivalent radius of the fire source, and real-time wind speed and direction data, the trajectory of the water flow in the air can be calculated. For the three bidirectional oscillating nozzles located at both ends and the middle of the fire extinguishing truss, the optimal horizontal oscillation angle and pitch angle are solved, improving the accuracy of the spray and avoiding blind sweeping. Meanwhile, by generating periodic oscillation patterns for each nozzle and dynamically adjusting the oscillation amplitude and frequency according to the size of the fire source, three-dimensional cross-coverage of the fire source area is achieved, improving the fire source coverage efficiency and the overall effectiveness of fire extinguishing operations.

[0051] In the description of this invention, it should be understood that the terms "longitudinal", "lateral", "up", "down", "front", "rear", "left", "right", "vertical", "horizontal", "top", "bottom", "inner", "outer", etc., indicate the orientation or positional relationship based on the orientation or positional relationship shown in the accompanying drawings, and are only for the convenience of describing this invention, and are not intended to indicate or imply that the device or element referred to must have a specific orientation, or be constructed and operated in a specific orientation, and therefore should not be construed as a limitation of this invention.

[0052] The above embodiments are merely descriptions of preferred embodiments of the present invention and are not intended to limit the scope of the present invention. Various modifications and improvements made by those skilled in the art to the technical solutions of the present invention without departing from the spirit of the present invention should fall within the protection scope defined by the claims of the present invention.

Claims

1. A suspended truss-type fire extinguishing device for ultra-high-rise buildings, characterized in that, include: Fire extinguishing mechanism (1) and control module, wherein the control module is electrically connected to the fire extinguishing mechanism (1) and the control module is used to control the fire extinguishing mechanism (1); The fire extinguishing mechanism (1) includes: The system includes a hoisting component, a fire extinguishing component, a terahertz transmitter, a temperature sensor, a smoke sensor, a wind speed sensor, and a drive component. The hoisting component is fixedly connected to the roof of the building, the fire extinguishing component is fixedly connected to the hoisting component, and the drive component is fixedly connected to both the hoisting component and the fire extinguishing component. The terahertz transmitter, the temperature sensor, the smoke sensor, and the wind speed sensor are all located on the fire extinguishing component. The control module includes: The sensing unit is configured to acquire terahertz image data, target distance data and ambient temperature data of the fire area based on the terahertz transmitter, and combine them with the real-time pose data of the fire extinguishing component to generate a sparse sensing dataset with spatiotemporal labels. The fire source inversion unit is configured to receive the sparse sensing dataset, construct a physical information neural network, use the fire source center coordinates and heat release rate as trainable variables of the network, and invert the fire source parameters by minimizing the total loss function that includes data fitting loss terms and physical equation residual loss terms. The water spray control unit is configured to receive the fire source parameters, establish a water flow trajectory model in the air, calculate the optimal spray angle of multiple water nozzles on the fire extinguishing truss based on the fire source parameters, generate a periodic oscillation pattern of each water nozzle, and convert the optimal spray angle and periodic oscillation pattern into control signal output. The early warning unit is configured to monitor the status of the fire source in real time during the water spraying process and determine whether the fire source has been extinguished.

2. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 1, characterized in that, The hoisting components include a track-mounted unmanned vehicle, a winch, and a telescopic boom. The track-mounted unmanned vehicle is fixedly connected to the roof of the building. The winch is mounted on the track-mounted unmanned vehicle and connected to the fire extinguishing components via the telescopic boom. The winch is used to drive the telescopic boom to move, and the telescopic boom is fixedly connected to the track-mounted unmanned vehicle.

3. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 2, characterized in that, The fire extinguishing component includes a truss body, several bidirectional swing water nozzles, and a water supply pipeline. The truss body is fixedly connected to the winch via the telescopic boom. Several bidirectional swing water nozzles are provided on the truss body. One end of the water supply pipeline is connected to a fire water tank, and the other end of the water supply pipeline is connected to the truss body.

4. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 3, characterized in that, When generating a sparse-aware dataset with spatiotemporal labels, the following steps are included: The three-dimensional position coordinates, pitch angle, roll angle and yaw angle of the fire extinguishing truss are acquired in real time as the real-time pose data; Acquire terahertz reflection intensity images and obtain target distance values ​​by phase method; Ambient temperature data is collected by an array of temperature sensors integrated on the fire extinguishing truss; The terahertz image data, distance value, and ambient temperature data of each scanning point are spatiotemporally aligned with the truss pose data at the corresponding time. Based on the truss pose and the installation position and orientation of the terahertz transmitter, calculate the spatial coordinates of each scanning point in the building coordinate system; Generate a sparse sensing dataset containing timestamps, spatial coordinates, terahertz image features, distance values, and ambient temperature values.

5. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 4, characterized in that, When obtaining the ignition source parameters through inversion, the following are included: A physical information neural network is constructed with spatial coordinates and time coordinates as inputs. The output of the physical information neural network includes temperature, pressure, velocity vector and smoke concentration. The fire dynamics physical equations, including the discretized forms of the Navier-Stokes equations, energy equations, and component transport equations, are embedded into the neural network as physical residual outputs. Construct a total loss function, which includes a data fitting loss term, a physical equation residual loss term, and a boundary condition loss term; The coordinates of the fire source center and the heat release rate are initialized as trainable variables, and gradient optimization is performed simultaneously with the weights of the physical information neural network to minimize the total loss function. Output the fire source parameters obtained by inversion, including the three-dimensional coordinates of the fire source center, the heat release rate, and the equivalent radius of the fire source.

6. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 5, characterized in that, When minimizing the total loss function, which includes both the data fitting loss term and the physical equation residual loss term, the following is included: The data fitting loss term is constructed, which includes at least a terahertz image loss term to constrain the temperature field output by the physical information neural network to be consistent with the high temperature region identified by the terahertz image at the measurement point; a distance loss term to constrain the fire source location predicted by the physical information neural network to be consistent with the distance value obtained by terahertz ranging; and an ambient temperature loss term to constrain the ambient temperature output by the physical information neural network to be consistent with the measured ambient temperature value. Construct the residual loss term of the physical equation, which is the sum of the residuals calculated after substituting the output of the physical information neural network into each physical equation; The boundary condition loss term is constructed as a constraint applied based on the positions of walls, doors, and windows in the building information model.

7. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 6, characterized in that, When receiving the fire source parameters and establishing a model of the water flow trajectory in the air, the process includes: Obtain water supply pressure parameters and determine water outlet velocity; Real-time wind speed and direction data are obtained by wind speed and direction sensors integrated on the fire extinguishing truss. Based on the water flow outlet velocity, wind speed and direction, and gravitational acceleration, a differential equation or analytical expression for the water flow trajectory is established as the motion trajectory model.

8. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 7, characterized in that, When calculating the optimal spray angle of multiple nozzles on the fire extinguishing truss based on the fire source parameters, the following is included: Based on the three-dimensional coordinates of the fire source center and the equivalent radius of the fire source output by the fire source inversion module, and combined with the motion trajectory model, the horizontal swing angle and pitch angle of multiple bidirectional swing nozzles on the fire extinguishing truss are respectively solved, wherein the multiple bidirectional swing nozzles are located at both ends and the middle of the fire extinguishing truss. The swing amplitude of each nozzle is determined based on the equivalent radius of the fire source, and the swing frequency of each nozzle is determined based on the fire intensity. Control parameters are generated to make each nozzle swing periodically in the horizontal and vertical directions. The flow rate of the three water nozzles is dynamically allocated according to the fire source parameters, with the two end nozzles focusing on covering both sides of the fire source and the middle nozzle focusing on covering the center of the fire source.

9. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 8, characterized in that, The water spray control unit is also configured to: during the water spraying process, acquire terahertz images and distance data of the fire source area in real time through the terahertz sensing and positioning module, and monitor the shape and temperature changes of the fire source; When the rate of temperature drop in the fire source area is detected to be lower than the preset threshold or a temperature rebound occurs, the injection parameters are recalculated or the swing strategy is adjusted.

10. The suspended truss-type fire extinguishing equipment for super high-rise buildings according to claim 9, characterized in that, Determining whether a fire source has been extinguished includes: Real-time acquisition of terahertz image sequences and distance data sequences of the fire source area; Extract the area, highest temperature value, and temperature distribution characteristics of the high-temperature region from the terahertz image sequence; When the area of ​​the high-temperature region is continuously smaller than the first preset threshold, the highest temperature value is continuously lower than the second preset threshold, and the temperature distribution shows a uniform downward trend, it is determined that the fire source has been extinguished.

Images