Dynamic temperature field real-time reconstruction method and system based on multi-source sound wave data

Abstract

The invention discloses a dynamic temperature field real-time reconstruction method and system based on multi-source sound wave data, and relates to the technical field of fire monitoring and early warning, and the method comprises the steps: deploying a sound wave receiving and transmitting array in a target region, constructing a multi-source sound wave measurement network, and determining a signal track space; connecting a visual simulation platform, constructing a verification database and storing the verification database to a platform data center; developing a reconstruction unit in the visual simulation platform, and determining a reconstruction temperature field; and executing linear measurement and calculation based on acoustic trajectory points and interpolation compensation of a non-trajectory region by taking a preset distributed interval as a constraint. According to the invention, the technical problems that the measurement is easy to interfere and the internal temperature is difficult to accurately acquire in real time in the high-temperature and dynamically-changing complex environment in the traditional method are solved, and the technical effect of accurately acquiring the internal temperature field information in real time in a non-contact manner in the high-temperature and dynamically-changing complex environment is achieved.

Description

Technical Field

[0001] The present invention relates to the technical field of fire monitoring and early warning, and particularly relates to a method and system for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data. Background Art

[0002] In the process of temperature field monitoring, traditional temperature measurement methods are widely used. For example, thermocouples, infrared thermometers, etc. are often used for temperature measurement in various scenarios. In a relatively stable environment, these methods can play a certain role. However, with the development of industry and the improvement of safety requirements, the requirement for accurate monitoring of temperature fields in complex environments is getting higher and higher. The disadvantages of traditional methods gradually emerge when facing complex environments with high temperature and dynamic changes. Contact measurement will interfere with the object to be measured and affect the measurement accuracy; non-contact measurement is difficult to obtain the internal temperature distribution and has poor real-time performance. Just like in a complex environment such as a fire scene, traditional measurement methods cannot obtain temperature field information in a timely and accurate manner, and it is difficult to meet the precise requirements for temperature field data in fire early warning, fire fighting and rescue decision-making, etc. Summary of the Invention

[0003] This application solves the technical problems that traditional methods are easily interfered during measurement and it is difficult to accurately and real-time obtain the internal temperature when facing complex environments with high temperature and dynamic changes. This application collects acoustic signal data by deploying an acoustic wave transceiver array inside the target area to construct a multi-source acoustic wave measurement network, determines the signal trajectory space, conducts fire dynamics simulation to construct a verification database, develops a reconstruction unit, and combines linear calculation and K-nearest neighbor interpolation and other processes to obtain temperature field-related data, and determines the reconstructed temperature field through dynamic evolution prediction, so as to accurately measure the dynamic temperature field situation of the target area in complex environments (such as fire scenes), making the temperature field reconstruction result more accurate, real-time and reliable.

[0004] In view of the above technical problems, this application proposes a technical solution for a method and system for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data.

[0005] In a first aspect, the present application provides a method for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data. The method includes: deploying an acoustic wave transceiver array inside a target area to construct a multi-source acoustic wave measurement network, triggering the co-frequency transmission and echo reception of acoustic signals to determine the signal trajectory space, where each acoustic wave transceiver is marked with a spatial position code; connecting to a visualization simulation platform to perform fire dynamics simulations under multiple fire conditions for the target area, constructing a verification database and storing it in the platform data center; developing a reconstruction unit in the visualization simulation platform based on the linear relationship between sound speed and temperature constrained by acoustic signal parameters, assisting the verification database, performing real-time analysis and dynamic evolution prediction based on linear measurement and K-nearest neighbor interpolation for the signal trajectory space to determine the reconstructed temperature field; performing linear measurement based on acoustic trajectory points and interpolation compensation for non-trajectory areas with a preset distributed spacing as a constraint.

[0006] In a second aspect, the present application provides a system for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data. The system includes: a signal trajectory space determination module for deploying an acoustic wave transceiver array inside a target area to construct a multi-source acoustic wave measurement network, triggering the co-frequency transmission and echo reception of acoustic signals to determine the signal trajectory space, where each acoustic wave transceiver is marked with a spatial position code; a platform data center storage module for connecting to a visualization simulation platform to perform fire dynamics simulations under multiple fire conditions for the target area, constructing a verification database and storing it in the platform data center; a reconstructed temperature field determination module for developing a reconstruction unit in the visualization simulation platform based on the linear relationship between sound speed and temperature constrained by acoustic signal parameters, assisting the verification database, performing real-time analysis and dynamic evolution prediction based on linear measurement and K-nearest neighbor interpolation for the signal trajectory space to determine the reconstructed temperature field; a linear measurement execution module for performing linear measurement based on acoustic trajectory points and interpolation compensation for non-trajectory areas with a preset distributed spacing as a constraint.

[0007] The present application proposes one or more technical solutions, having at least the following technical effects:

[0008] In this application, an acoustic wave transceiver array is deployed within the target area to clarify the measurement area and the data acquisition object. Then, the synchronous transmission of acoustic signals and the reception of echoes are triggered, and the signal trajectory space is determined by analyzing the echoes to determine the transmission trajectory and the spatial distribution trajectory. Next, a visualization simulation platform is connected to determine the dynamic elements and cluster the fire scenarios, simulate the fire dynamics, and construct a verification database. After that, a multi-head self-attention target is introduced into the platform to construct a prediction and reconstruction area, which is combined with the real-time reconstruction area to form a reconstruction unit. Using this unit, the static temperature field is determined through linear measurement and K-nearest neighbor interpolation, and then the reconstructed temperature field is obtained through dynamic evolution prediction. At the same time, the connection between the acoustic wave transceiver array, the simulation platform, and the fire monitoring system is established to achieve front-end detection, simulation reconstruction decision-making, and response alarm, achieving the technical effect of non-contact, accurate, and real-time acquisition of internal temperature field information in a high-temperature and dynamically changing complex environment.

[0009] The above content outlines this application's method and system for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data. The technical solution steps of this application will be described in detail in the following specific embodiments to facilitate a clear and complete understanding of this application by those skilled in the art. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] To more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention, and those of ordinary skill in the art can obtain other drawings without creative efforts based on these drawings.

[0011] Figure 1 It is a flowchart of the method for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data provided by an embodiment of this application.

[0012] Figure 2 It is a schematic structural diagram of the system for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data provided by an embodiment of this application.

[0013] Figure 3 It is a schematic diagram of the deployment of the building space experimental device for the method for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data provided by an embodiment of this application.

[0014] Figure 4 It is a schematic diagram of the distribution of the effective acoustic wave paths of the building space experimental device for the method for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data provided by an embodiment of this application.

[0015] Figure 5 It is a schematic diagram of the distribution of the effective acoustic wave paths and the temperature points on the paths in the temperature field for the method for real-time reconstruction of a dynamic temperature field based on multi-source acoustic wave data provided by an embodiment of this application.

[0016] Figure 6 It is a schematic diagram of a set of visualization images of a sonic device for the real-time reconstruction method of a dynamic temperature field based on multi-source sonic data provided by an embodiment of the present application.

[0017] Explanation of reference numerals: Signal trajectory space determination module 1, platform data center storage module 2, reconstructed temperature field determination module 3, linear measurement execution module 4, sound wave generator 11, ignition point 12. Detailed implementation manners

[0018] In the present application, a sonic transceiver array is deployed in a target area to construct a multi-source sonic measurement network, trigger the emission and reception of sound signals, analyze the echoes to determine the signal transmission trajectory and construct a signal trajectory space. Connect to a visualization simulation platform, determine dynamic elements, cluster fire scenarios, and simulate fire dynamics to construct a verification database. Introduce a multi-head self-attention target in the platform to construct a prediction reconstruction area, and combine it with the real-time reconstruction area to form a reconstruction unit. Use the reconstruction unit to determine the static temperature field through linear measurement and K-nearest neighbor interpolation, and then obtain the reconstructed temperature field through dynamic evolution prediction. Establish connections between various systems to achieve front-end detection, simulation reconstruction decision-making, and alarm response, achieving the technical effect of non-contact, accurate, and real-time acquisition of internal temperature field information in a high-temperature and dynamically changing complex environment.

[0019] Next, the technical solutions in the embodiments of the present application will be clearly and completely described with reference to the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0020] It should be noted that any variations of the terms "include" and "have" are intended to cover non-exclusive inclusions. For example, a process, method, system, product, or server that includes a series of steps or units does not necessarily have to be limited to those clearly listed steps or units, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products, or devices.

[0021] Embodiment 1, as shown in Figure 1 , Figure 3 , Figure 4 , Figure 5 and Figure 6 shown, for the real-time reconstruction method of a dynamic temperature field based on multi-source sonic data, wherein the method includes:

[0022] Step A1000: Deploy an acoustic wave transceiver array inside the target area, construct a multi-source acoustic wave measurement network, trigger the co-frequency transmission and echo reception of acoustic signals, and determine the signal trajectory space. Among them, each acoustic wave transceiver is marked with a spatial position code.

[0023] In the embodiments of the present application, the acoustic wave transceiver array is a combination of devices deployed in the target area for transmitting and receiving acoustic signals. The multi-source acoustic wave measurement network is constructed by an acoustic wave transceiver array with spatial position codes deployed in the target area. By triggering the co-frequency transmission and echo reception of acoustic signals, data is obtained from multiple sound sources and receiving points, providing an acoustic data basis for the real-time reconstruction of the dynamic temperature field. The signal trajectory space is a space formed by receiving and analyzing the echo signals to determine the signal transmission trajectories marked with the time-of-flight, constructing a three-dimensional space with the target area, and distributing these signal transmission trajectories in this three-dimensional space. The spatial position code is the code assigned to each acoustic wave transceiver in the acoustic wave transceiver array, used to identify its position in space.

[0024] Specifically, inside the target area (the area where real-time reconstruction of the dynamic temperature field is required), first deploy the acoustic wave transceiver array, and assign spatial position codes to each acoustic wave transceiver and the ignition point in the area. For example, Figure 3 as shown by acoustic wave device 11 and ignition point 12, Figure 6 shows a set of visualization images of acoustic wave devices, which includes the distribution of 16 acoustic wave devices in space. For example, in a target area with a length, width, and height of 10 meters, 10 meters, and 10 meters respectively, a three-dimensional coordinate system is established with the lower left corner of the area as the origin, and each transceiver has corresponding (x, y, z) coordinates as its spatial position code, accurate to 0.1 meter.

[0025] After the deployment is completed, determine numerous signal transmission trajectories, construct a three-dimensional space with the target area, and distribute these trajectories in the three-dimensional space, thereby determining the signal trajectory space. The specific steps are described in detail in A1010 - A1020.

[0026] Determining the signal trajectory space provides a key data basis for subsequent temperature field reconstruction based on acoustic wave data, enabling more accurate analysis of the propagation of acoustic waves in the target area and laying an important foundation for realizing the real-time reconstruction of the dynamic temperature field.

[0027] Step A2000: Connect to the visualization simulation platform, perform fire dynamics simulation under multiple fire conditions for the target area, construct a verification database and store it in the platform data center.

[0028] In the embodiments of the present application, the visual simulation platform integrates various functions such as simulating fire scenes, processing data, and implementing temperature field reconstruction algorithms. It is the core platform for data processing and analysis in the entire method process. Multiple fire conditions are determined by clustering the fire scenes in the target area according to dynamic elements (including fire sources, air flow, thermal radiation, natural convection, etc.). Fire dynamics simulation takes the dynamic evolution of distributed temperature as the simulation target. On the visual simulation platform, for multiple fire conditions, relevant algorithms and models are used to simulate the development process of the fire in the target area. The verification database is a database constructed by integrating the fire laws obtained from fire dynamics simulations under various fire conditions. The platform data center is the place where the visual simulation platform stores data.

[0029] Optionally, first, ensure that the software and hardware environment of the visual simulation platform is complete, establish a stable physical connection channel (for example, use Gigabit Ethernet connection to meet the need for rapid transmission of a large amount of simulation data), perform software docking and call the API (Application Programming Interface) or specific connection tools provided by the platform, and configure parameters according to its specifications. These parameters include basic information such as the spatial dimensions and coordinate system of the target area. If compatibility issues are encountered during the connection, update the driver or contact technical support. After successful connection, the platform receives data and conducts simulations.

[0030] Next, determine the dynamic elements, which at least include fire sources, air flow, thermal radiation, natural convection, etc.; according to these dynamic elements, cluster the fire scenes in the target area to determine multiple fire conditions; conduct fire dynamics simulations on the visual simulation platform for these multiple fire conditions; finally, integrate the simulation results under each fire condition, that is, integrate information such as temperature change data and fire spread conditions obtained under different fire scenes, construct a verification database, and store it in the platform data center. The specific steps are described in detail in A2010 - A2040.

[0031] By constructing the verification database through the above steps, it provides an important reference basis for the subsequent real - time reconstruction of the dynamic temperature field based on multi - source acoustic data, and can help improve the accuracy and reliability of temperature field reconstruction.

[0032] Step A3000: Develop a reconstruction unit in the visual simulation platform based on the linear relationship between sound speed and temperature under the constraint of sound signal parameters, assist the verification database, perform real - time analysis based on linear measurement and K - nearest neighbor interpolation for the signal trajectory space, and dynamic evolution prediction to determine the reconstructed temperature field.

[0033] In the embodiments of the present application, the acoustic signal parameter constraint refers to relevant parameters of the acoustic signal, such as frequency, amplitude, flight time, etc. The development and reconstruction unit introduces multi-head self-attention targets aiming at global, local, and dynamic features, constructs parallel branches to form a prediction and reconstruction area and combines it with the real-time reconstruction area, and assists the verification database to perform real-time analysis and dynamic evolution prediction on the signal trajectory space, and couples the branch output results as the reconstruction result. Reconstructing the temperature field is a process of using linear measurement and K-nearest neighbor interpolation for real-time analysis, performing dynamic evolution prediction and integrating the results to achieve real-time and accurate presentation of the temperature field in the target area.

[0034] In an embodiment of the present application, first, develop a reconstruction unit in the visual simulation platform. First, introduce multi-head self-attention targets aiming at global, local, and dynamic features, then construct multiple parallel branches based on this and integrate them in parallel as the prediction and reconstruction area, and finally use the real-time reconstruction area and the prediction and reconstruction area as the reconstruction unit. Each branch performs directional decision-making and dynamic evolution prediction, and couples the branch output results to obtain the reconstruction result. The specific steps are described in detail in A3010 - A3030.

[0035] Next, use the verification database to assist the reconstruction unit in working. When performing real-time analysis, perform linear measurement on the signal trajectory space. First, identify the signal trajectory space based on the real-time reconstruction area, determine effective trajectory nodes according to the preset distributed spacing, then detect the acoustic wave parameters of these nodes, match the linear relationship between sound speed and temperature to obtain the target linear relationship, and finally use this relationship to calculate the temperature values of the effective trajectory nodes. The specific steps are described in detail in A3040 - A3060.

[0036] For non-trajectory areas, perform K-nearest neighbor interpolation. First, locate distribution points in the non-trajectory area at the preset distributed spacing, then use the K-nearest neighbor interpolation method to calculate the temperature values of each distribution point based on the distance-weighted calculation of the K nearest neighbor points of the distribution points, and finally construct a static temperature field based on these temperature values. The specific steps are described in detail in A3070 - A3090.

[0037] Finally, perform dynamic evolution prediction. For the constructed static temperature field, perform feature directional evolution decision-making under multiple parallel branches to determine the branch prediction results, then couple these branch prediction results, and integrate them with the time series and the static temperature field to finally determine the reconstructed temperature field. The specific steps are described in detail in A3100 - A3110.

[0038] By continuously updating data and repeating the above steps, the real-time reconstruction of the dynamic temperature field is realized, providing accurate temperature field information support for fire monitoring, prevention, and control, etc.

[0039] Step A4000: Among them, with the preset distributed spacing as a constraint, perform linear measurement based on acoustic trajectory points and interpolation compensation for non-trajectory areas.

[0040] In the embodiments of the present application, the distributed spacing is a distance standard for constraining linear measurement and interpolation compensation of non-trajectory regions. The acoustic trajectory points refer to the points on the acoustic wave signal transmission trajectory that are closely related to determining the signal trajectory space and performing linear measurement.

[0041] Specifically, after determining the signal trajectory space, perform linear measurement based on the acoustic trajectory points with reference to a preset distributed spacing. With the preset distributed spacing (for example, the set spacing is 1 meter) as a constraint, identify the signal trajectory space based on the real-time reconstruction area, and mark the effective trajectory nodes of the signal transmission trajectory. For example, on the signal transmission trajectory, determine an effective trajectory node every 1 meter. For these effective trajectory nodes, detect acoustic wave parameters (such as the propagation time, frequency, etc. of the acoustic wave), and use the linear relationship between sound speed and temperature (it is known that in a specific environment, the sound speed and temperature satisfy a specific linear formula, such as v = 331.4 + 0.6T, where v is the sound speed and T is the temperature) for matching to determine the target linear relationship. In this way, the temperature values of the effective trajectory nodes are calculated.

[0042] For non-trajectory regions, perform interpolation compensation. Also with the preset distributed spacing as a constraint, locate the distribution points of the non-trajectory regions. For example, in the areas of a warehouse where there is no direct propagation of acoustic waves, determine the distribution points at an interval of 1 meter. For these distribution points, use the K-nearest neighbor interpolation method to calculate the temperature values. Assume that the value of K is taken as 5, that is, find the 5 points closest to each distribution point (these points can be effective trajectory nodes or other points for which the temperature values have been calculated), and calculate the temperature value of this distribution point based on the distance-based weighting of these 5 nearest neighbor points. The closer the point, the greater its weight. For example, if the temperatures of the 5 nearest neighbor points around a certain distribution point are T1, T2, T3, T4, T5, and their distances from this distribution point are d1, d2, d3, d4, d5 respectively, then the temperature value T of this distribution point = (w1T1 + w2T2 + w3T3 + w4T4 + w5T5) / (w1 + w2 + w3 + w4 + w5), where the weights w1 - w5 are determined according to the distance. Based on the temperature values of each distribution point, construct a static temperature field.

[0043] Through the above steps of linear measurement based on acoustic trajectory points and interpolation compensation of non-trajectory regions, the temperature information of each position in the target area can be obtained more accurately, realizing a more accurate reconstruction of the temperature field in the target area, and providing strong support for application scenarios such as fire monitoring and temperature control in industrial production processes.

[0044] Furthermore, step A1000 in the method provided by the embodiments of the present application includes:

[0045] A1010: Receive and parse the echo signal, and determine the signal transmission trajectory, where each signal transmission trajectory is marked with the time of flight.

[0046] A1020: Construct a three-dimensional space with the target area, distribute the signal transmission trajectories in space, and determine the signal trajectory space.

[0047] In the embodiments of the present application, the signal transmission trajectory refers to the propagation path of the acoustic signal from the transmitter through the target area and then received by the receiver. The time of flight refers to the duration from the transmission to the reception of the acoustic signal, which is marked on each signal transmission trajectory.

[0048] Specifically, first, after deploying the acoustic wave transceiver array inside the target area, the array will trigger the co-frequency transmission of acoustic signals. Trigger the acoustic wave transceiver array to perform co-frequency transmission and echo reception of acoustic signals. Assume that the frequency of the transmitted acoustic signal is 50 kHz (this frequency can propagate stably in the target area and is not easily interfered). When the acoustic signal is transmitted and propagates in the target area, it will reflect back to form an echo when encountering an obstacle. When receiving the echo, record the reception time of the echo. By calculating the difference between the transmission time and the reception time, the time of flight of the signal can be obtained. For example, if the transmission time is t1 = 0.001 s and the reception time is t2 = 0.003 s, then the time of flight Δt = t2 - t1 = 0.002 s. Given that in the air medium of this target area, the speed of sound v = 340 m / s (obtained through experiments or theoretical calculations by those skilled in the art), according to the formula s = v×Δt, the propagation distance of the signal s = 340×0.002 = 0.68 m can be calculated.

[0049] Since each acoustic wave transceiver has a spatial position code, by combining the echo information received by multiple transceivers and their spatial positions, the signal transmission trajectory can be determined using the triangulation method. For example, there are three transceivers A(1.0, 2.0, 3.0), B(3.0, 4.0, 5.0), C(5.0, 6.0, 7.0), which receive the same echo respectively. The propagation distances calculated according to their times of flight are sA = 0.5 m, sB = 0.7 m, sC = 0.9 m. By solving the system of equations of the distance formula between two points in space and solving them simultaneously, the corresponding x, y, z values are obtained. The coordinate points determined by these values are the positions of the signal source, thereby determining the propagation path of the signal in space, and the signal transmission trajectory in space can be accurately determined.

[0050] Next, a three-dimensional space is constructed based on the target area to spatially distribute the signal transmission trajectories, thereby determining the signal trajectory space. Taking a three-dimensional space with dimensions of 10×10×10 in length, width, and height as an example, the spatial position encoding of the acoustic wave transceiver is used as a reference for the spatial coordinates. According to the propagation characteristics of the acoustic signal and the flight time, by calculating the product of the sound speed and the flight time, the distance traveled by the acoustic signal on each transmission trajectory can be obtained. Given that in this factory environment, the sound speed is approximately 340 m / s, and combined with the previously mentioned flight time of 0.002 s, the propagation distance of this signal transmission trajectory can be calculated as 0.68 m. Based on this distance information and the positions of the acoustic wave transceivers, the signal transmission trajectories can be accurately distributed in the constructed three-dimensional space. After all the signal transmission trajectories are distributed in the three-dimensional space, the spatial range occupied by these trajectories constitutes the signal trajectory space.

[0051] By determining the signal trajectory space, it provides a key spatial information basis for the subsequent real-time reconstruction of the dynamic temperature field based on multi-source acoustic wave data, which helps to more accurately analyze the temperature distribution in the target area.

[0052] Furthermore, step A2000 in the method provided by the embodiments of the present application includes:

[0053] A2010: Determine the dynamic elements, where the dynamic elements at least include a fire source, air flow, thermal radiation, and natural convection.

[0054] A2020: Cluster the fire scenarios in the target area according to the determined dynamic elements to determine multiple fire conditions.

[0055] A2030: Conduct fire dynamics simulations for the multiple fire conditions to determine the fire behavior patterns, where the dynamic evolution of the distributed temperature is taken as the simulation target.

[0056] A2040: Integrate the fire behavior patterns under each fire condition to construct the verification database.

[0057] In the embodiments of the present application, the fire scenario clustering is a process of classifying the fire scenarios in the target area according to the determined dynamic elements.

[0058] Optionally, first, determine the dynamic elements. These elements such as the fire source, air flow, thermal radiation, and natural convection play a key role in the development of the fire and the change of the temperature field. Assume that a fire occurs in an indoor space (target area) with a length of 10 m, a width of 10 m, and a height of 10 m. The fire source is set as a combustion source with a power of 500 kW, located at the center of the room; the air flow may be caused by the ventilation system, and the wind speed is set to 1.5 m / s, with the wind direction blowing from one side to the other.

[0059] In terms of thermal radiation, based on the fire source power and the thermal radiation characteristics of the surrounding materials, assuming the emissivity of the surrounding wall material is 0.8, and the measured wall temperature is t = 300 °C. Using the Stefan-Boltzmann law J = εσT 4 to calculate the thermal radiation intensity, where J is the thermal radiation intensity (unit: W / m 2 ), ε is the emissivity of the object surface, σ is the Stefan-Boltzmann constant, σ = 5.67×10 -8 W / (m 2 ·K 4 ), and T is the absolute temperature of the object (unit: K). First, convert the wall temperature to absolute temperature T = t + 273.15 = 300 + 273.15 = 573.15 K. Substitute the above ε, σ, and T into the formula J = εσT 4 to obtain the thermal radiation J = 0.8×5.67×10 -8 ×(573.15) 4 .

[0060] For natural convection, based on the temperature difference in the indoor space and the physical properties of the air, calculate its intensity through relevant formulas:

[0061] Step a: Substitute the measured or known indoor temperature and air physical property data into the formula to calculate the Rayleigh number (the Rayleigh number comprehensively reflects the interaction of buoyancy, viscous force, and heat conduction in natural convection). g is the acceleration due to gravity (usually taken as 9.8 m / s2), α is the thermal expansion coefficient of the air (at normal temperature and pressure, the thermal expansion coefficient of the air T is the absolute temperature of the air, unit: K), ΔT is the temperature difference in the indoor space (for example, the temperature difference between the high-temperature area near the fire source and the area far from the fire source), L is the characteristic length (which can be the height of the room, the size of the object, etc.), ν is the kinematic viscosity of the air (at normal temperature and pressure, the kinematic viscosity of the air is about 1.5×10 -5 m 2 / s), α th is the thermal diffusivity of the air (at normal temperature and pressure, the thermal diffusivity of the air is about 2.2×1010 -5 m 2 / s).

[0062] Step b: Determine the flow state according to the Rayleigh number calculated in step a. Those skilled in the art select appropriate constants and substitute them into the correlation formula to calculate the Nusselt number (the Nusselt number characterizes the strength of convective heat transfer, and there are different correlation formulas for different natural convection scenarios). In the case of large-space natural convection such as indoor natural convection, a common correlation formula is Nu = CRa n, C and n are constants, depending on factors such as the flow state of natural convection and the shape of the object. For natural convection in a large space and in the laminar state (generally Ra < 10 9 ), C = 0.59 and n = 1 / 4; for the turbulent state (Ra > 10 9 ), C = 0.1 and n = 1 / 3.

[0063] Step c: According to the definition of the Nusselt number the natural convection heat transfer coefficient can be deduced inversely where λ is the thermal conductivity of air (at normal temperature and pressure, the thermal conductivity of air is about 0.026 W / (m·K)), and L is the characteristic length mentioned in step a. Substitute the calculated Nusselt number and the known thermal conductivity of air and characteristic length into it to obtain the natural convection heat transfer coefficient.

[0064] Step d: The natural convection intensity is generally represented by the natural convection heat transfer quantity Q. According to Newton's cooling formula Q = hAΔT, A is the heat transfer area (such as the contact area between the wall and air, the contact area between the fire source and air, etc.), h is the natural convection heat transfer coefficient calculated previously, and ΔT is the temperature difference used when calculating the Rayleigh number previously. Substitute these data into the formula to calculate the natural convection intensity.

[0065] Next, cluster the fire scene in the target area according to the determined dynamic elements to determine multi-fire conditions. Collect a large amount of relevant data according to the above method for determining dynamic elements. After collecting the data, use the K-Means clustering algorithm to process these data:

[0066] Step e: Determine the number of clusters, that is, the number of fire scene categories to be divided. This number can be determined according to the actual experience of those skilled in the art or through multiple experiments. Assume that the number of clusters is determined to be 5, and the algorithm will randomly select 5 initial cluster centers, which represent different combinations of fire scene characteristics.

[0067] Step f: Calculate the distance from each data point to these 5 cluster centers. Here, the distance can be calculated using the Euclidean distance calculation method. According to the distance, allocate the data points to the category where the nearest cluster center is located.

[0068] Step g: After the allocation is completed, recalculate the cluster center of each category, that is, the average value of all data points in each category in the dimension of each dynamic element. Keep repeating this process until the cluster center no longer changes significantly or changes very little. At this time, the clustering is completed.

[0069] After the above clustering process, data points with combined characteristics such as similar fire source power, air flow velocity and direction, heat radiation intensity, and natural convection intensity can be grouped into one category, thereby determining various different fire scenario categories such as high-power fire source - strong air flow, low-power fire source - weak air flow.

[0070] Then, for these multiple fire conditions, simulations are carried out on a professional fire dynamics simulation software platform (such as FDS, Fire Dynamics Simulator). The simulation takes the dynamic evolution of the distributed temperature as the core goal. The target area is divided into numerous tiny grid cells, and the size of each grid cell is set according to the simulation accuracy requirements, assumed to be 0.5 m × 0.5 m × 0.5 m. As the simulation time progresses, the temperature change of each grid cell is recorded at regular time intervals (assumed to be 1 second). For example, when simulating the scenario of high-power fire source - strong air flow, it can be observed that the temperature of the grid cells near the fire source rises rapidly, and under the action of the air flow, the high-temperature area spreads along the air flow direction, and the temperature distribution keeps changing. Through a large number of such simulations, the fire laws under different fire conditions can be determined.

[0071] Finally, integrate the fire laws under each fire condition to construct a verification database. Integrate information such as the temperature change data obtained from simulations under various fire scenarios, the fire spread path, and the temperature change curves of different regions over time to form a rich database. This verification database provides an important reference basis for the subsequent real-time reconstruction of the dynamic temperature field based on multi-source acoustic wave data. In practical applications, when real-time acoustic wave data is obtained, the temperature field can be reconstructed more accurately with the help of the verification database.

[0072] By constructing the verification database, it provides strong support for fire prevention, fire fighting, and the safe evacuation of personnel.

[0073] Furthermore, step A5000 in the method provided by the embodiments of the present application includes:

[0074] A5010: Establish connections among the acoustic wave transceiver array, the visualization simulation platform, and the fire monitoring system.

[0075] A5020: Among them, the acoustic wave transceiver array performs front-end detection, the reconstruction unit developed in the visualization simulation platform and the verification database perform simulation reconstruction decision-making, and the fire monitoring system performs response and alarm.

[0076] In the embodiments of the present application, front-end detection refers to deploying an acoustic wave transceiver array inside the target area, constructing a multi-source acoustic wave measurement network, triggering the co-frequency emission and echo reception of acoustic signals, and determining the signal trajectory space.

[0077] Specifically, first, deploy the acoustic wave transceiver array and perform front-end detection. The specific steps are detailed in A1000 and A1010 - A1020.

[0078] Next, connect to the visualization simulation platform. Transmit the data collected by the acoustic wave transceiver array to the visualization simulation platform, which conducts fire dynamics simulations under multiple fire conditions in the target area. The specific steps are detailed in A2010 - A2040.

[0079] Develop a reconstruction unit within the visualization simulation platform. When receiving the data from the acoustic wave transceiver array, the reconstruction unit combines with the verification database and performs real-time analysis based on linear measurement and K-nearest neighbor interpolation for the signal trajectory space to determine the reconstructed temperature field. The specific steps are detailed in A3010 - A3110.

[0080] The fire monitoring system establishes connections with the acoustic wave transceiver array and the visualization simulation platform. When the reconstruction unit within the visualization simulation platform determines the occurrence of a fire and obtains the temperature field information, the fire monitoring system receives this information and performs a response alarm. For example, when the reconstructed temperature field shows that the temperature in a certain area exceeds the set safety threshold (assumed to be 80°C), the fire monitoring system immediately issues an alarm to notify relevant personnel to take measures, achieving timely response and handling of the fire.

[0081] Through the above series of steps, the method for real-time reconstruction of the dynamic temperature field based on multi-source acoustic wave data can achieve real-time and accurate monitoring and alarm of the temperature field at the fire scene, providing strong support for fire prevention and control.

[0082] Furthermore, step A3000 in the method provided by the embodiments of the present application includes:

[0083] A3010: Introduce the multi-head self-attention target, where at least the global feature, local feature, and dynamic feature are targeted.

[0084] A3020: According to the multi-head self-attention target, construct multiple parallel branches and perform parallel integration as the predicted reconstruction area.

[0085] A3030: Use the real-time reconstruction area and the predicted reconstruction area as the reconstruction unit, where each branch performs directional decision-making and dynamic evolution prediction, and the results output by the branches are coupled as the reconstruction result.

[0086] In the embodiments of the present application, the multi-head self-attention target is the target introduced when developing the reconstruction unit within the visualization simulation platform. The parallel branches are multiple branches constructed according to the multi-head self-attention target. The predicted reconstruction area is composed of the parallel integration of multiple parallel branches and is used for predicting and reconstructing the temperature field.

[0087] Specifically, first, the multi-head self-attention objective is introduced with the aim of capturing multi-dimensional feature information. Taking a large industrial plant as an example, the global features can be obtained by collecting temperature data from various areas within the entire building, such as the overall temperature levels in different areas; the local features focus on the temperature changes of specific equipment or small-scale areas, such as the temperature fluctuations around a certain key machine; the dynamic features focus on the temperature changes over time, such as the rate of temperature rise during a fire.

[0088] Then, to achieve accurate modeling of the mapping relationship between multi-source acoustic wave data and the temperature field, under the framework of the multi-head self-attention objective, a deep learning model consisting of a series connection of an Adaptive Weighted Hybrid Convolution Network (AWHC) and a Dynamic Residual Attention-aware Fusion Network (DRAAFN) is constructed. The specific processes are as follows:

[0089] Design of the AWHC module: Three different convolution operations (3×3 standard convolution, 5×5 dilated convolution, global average pooling) are used to capture multi-scale features. The adaptive weighted layer assigns weights to the channel features to highlight the impact of key acoustic features (such as sound speed, time of flight) on temperature. Then, max pooling is used for dimensionality reduction to output a feature tensor that fuses local and global information.

[0090] Step h: Three parallel convolution branches are used to extract features. Branch 1: 3×3 standard convolution (stride 1, padding 1) to capture local details of acoustic wave propagation (such as local fluctuations in the time of flight of the signal); Branch 2: 5×5 dilated convolution (dilation rate 2, stride 1, padding 2) to expand the receptive field and fuse medium-scale sound speed distribution features; Branch 3: global average pooling to extract the global acoustic wave propagation trend (such as the regional average sound speed). Let the output features of the three convolution operations be H2 (local features), D2 (medium-scale features), and H d2 (global features). They are concatenated in the channel dimension to get C = Concatenate[H2, D2, H d2

[0091] Step i: The concatenated features are weighted by the adaptive weighted layer. A fully connected layer (number of neurons = number of channels after concatenation) and the softmax function are used to calculate the channel weights W = softmax(Dense(C)), f combined = C ⊙ W where ⊙ represents element-wise multiplication, W is obtained from the fully connected layer and the softmax function, and C is the total number of channels after concatenation.

[0092] Step j: Max pooling is performed on the weighted feature f combined for dimensionality reduction: f pooled = MaxPooling1D(2)(f combined ). ​

[0093] DRAAFN Module Design: Based on the AWHC output, a multi-attention mechanism and a residual structure are introduced. The global feature processing module is used to extract the macroscopic distribution trend of the temperature field, and the multi-branch residual structure (with different convolution kernel sizes) is used to capture local subtle temperature changes. The self-attention and channel attention are combined to dynamically adjust the feature weights. Finally, the global, local, and dynamic features are fused to output high-precision temperature prediction features.

[0094] Step k: Global feature processing. Let the global feature output by AWHC be denoted as f pooled . After operations such as convolution and activation, the tensor G2 is obtained, and then the final global feature representation G = Reshape(G2) is obtained through Reshape.

[0095] Step l: Local feature extraction. The input feature is convolved and then max-pooled to obtain the local feature L local = MaxPooling1D(L1).

[0096] Step m: Multi-branch residual structure. Parallel branches with different convolution kernel sizes are adopted to capture local information at different scales: B1 = σ(W b1 *L local +b b1 ), B2 = σ(W b2 *L local +b b2 ), where σ is the non-linear activation function, representing the convolution operation.

[0097] Step n: Dynamic attention mechanism, including two parts: self-attention and channel attention: A = MultiHeadAttention(h,d k )(L local ,L local ), where h is the number of attention heads, d k is the key vector dimension, and A and C a are the features output by self-attention and channel attention respectively.

[0098] Step n: Fusion and normalization. The global feature G, branch features B1, B2, and attention features A, C a are fused and layer normalization is performed: F m = Add([G,B1,B2,A,C a ), F o = LayerNormalization(F m ,ò), where ò is a very small positive number to prevent numerical instability.

[0099] Physical Constraint Joint Optimization Training Strategy: Design a multi-objective joint loss function that includes a temperature prediction error term (MAE) and an acoustic propagation physical constraint term to ensure that the predicted temperature field of the model not only conforms to the measured data but also satisfies the theoretical relationship between sound speed and temperature (e.g., v = 331.4 + 0.6T). At the same time, embed the smoothness constraint of the temperature field to suppress local abnormal fluctuations and improve the robustness of the model in complex environments.

[0100] Step o: Design of the multi-objective joint loss function, temperature prediction error term (MAE) where T n is the temperature predicted by the model, and is the temperature of the numerical simulation label. Acoustic propagation physical constraint term where is the flight time measured by reality or simulation, is then calculated by integrating the predicted temperature field of the model along the acoustic wave path: K is the number of sampling points discretized by the nth acoustic wave path, ||Δs k || is the distance between adjacent sampling points. The joint optimization objective function L total = λ · L MAE + (1 - λ) · L time , 0.6 ≤ λ ≤ 0.8.

[0101] Step p: Physical rule embedding mechanism, explicit constraint of the sound speed-temperature relationship, directly embed in the loss function or network calculation to ensure the consistency of the predicted temperature field in terms of acoustic propagation characteristics. Gradient backpropagation guidance: The physical constraint term corrects the gradients of the network parameters, guiding the model to satisfy physical laws such as the consistency of the acoustic wave flight time while fitting the temperature field.

[0102] Step q: Regularization enhancement strategy, smoothness constraint of the temperature field: Add a temperature gradient regularization term to the loss function to suppress local abnormal fluctuations of the predicted temperature. Expand the joint objective function L total = λL MAE + (1 - λ)L time + μL smooth , μ ∈ [0.005, 0.02].

[0103] Next, multiple parallel branches are constructed based on the multi-head self-attention target, similar to multi-threaded processing, and each branch focuses on different aspects of analysis and prediction. In actual operation, it is assumed that 5 parallel branches are constructed. Branch 1 focuses on analyzing the spatial distribution of temperature, branch 2 focuses on the time series characteristics of temperature changes, branch 3 focuses on the temperature change trend near the fire source, branch 4 focuses on the influence of vents on the temperature field, and branch 5 focuses on the temperature conduction characteristics of different material surfaces. Each branch independently analyzes the data and integrates them in parallel to form a prediction reconstruction area.

[0104] Data acquisition in the real-time reconstruction area relies on real-time analysis based on multi-source acoustic wave data. The data collected by the acoustic wave transceiver array is used to identify the valid trajectory nodes of the signal transmission trajectory according to the preset distributed spacing. Assume that at a certain moment, 100 valid trajectory nodes are determined in the factory building, and the acoustic wave parameters are detected for these nodes, and the linear relationship between sound speed and temperature is matched to calculate the temperature value of each node. The specific steps are described in detail in A3040-A3060. For non-trajectory areas, the distribution points are located and the temperature values ​​are calculated by K-nearest neighbor interpolation, so as to construct the real-time temperature field distribution. The specific steps are described in detail in A3070-A3090. For example, in a non-trajectory area, 50 distribution points are determined with a preset distributed spacing of 1 meter, and the temperature values ​​of these points are calculated by K-nearest neighbor interpolation (K is 5), so as to obtain the complete distribution of the real-time temperature field.

[0105] In the dynamic evolution prediction stage, each branch performs directional decision-making and dynamic evolution prediction. Taking branch 1 as an example, based on the temperature spatial distribution data of the real-time reconstruction area, combined with the simulation rules of similar temperature scenarios in the verification database, the diffusion direction and speed of temperature in the future period are predicted; branch 2 analyzes the periodicity and trend of temperature changes based on time series characteristics and historical data, and predicts the rhythm of temperature changes. The results of these branch outputs are not independent, but coupled. For example, when predicting the future temperature of a certain area, if 3 branches predict that the temperature of the area will rise, 1 branch predicts that it will remain unchanged, and 1 branch predicts that it will fall, according to the rule of "selecting the branch output with the highest proportion for the same target as the result of the target", the temperature of the area is determined to rise. In this way, the prediction results of multiple branches are integrated, and the time series and the static temperature field of the real-time reconstruction area are combined to finally determine the reconstructed temperature field.

[0106] Through the above methods, dynamic and accurate reconstruction of the temperature field is achieved, providing strong support for fire monitoring, indoor environment control, etc.

[0107] Further, step A3000 in the method provided in the embodiment of the present application includes:

[0108] A3040: Identify the signal trajectory space according to the real-time reconstruction area, and identify the effective trajectory nodes of the signal transmission trajectory at the preset distributed interval.

[0109] A3050: For the effective trajectory nodes, detect acoustic wave parameters for acoustic velocity-temperature linear relationship matching to determine the target linear relationship.

[0110] A3060: Calculate the temperature value of the effective trajectory nodes according to the target linear relationship.

[0111] In the embodiments of the present application, the effective trajectory nodes are key position points in the signal transmission trajectory that can reflect the acoustic wave propagation characteristics and temperature-related information.

[0112] Specifically, first, deploy an acoustic wave transceiver array in the target area to construct a multi-source acoustic wave measurement network (detailed in A1000). By triggering the same-frequency transmission and echo reception of acoustic signals, the signal trajectory space has been determined.

[0113] Based on this, use the real-time reconstruction area to identify the signal trajectory space. Suppose in an indoor space with an area of 100 square meters, based on the real-time reconstruction area data obtained through preliminary measurement and calculation, combined with the preset distributed interval (such as set to 1 meter), mark the effective trajectory nodes on the signal transmission trajectory, which is equivalent to marking key locations at a certain interval on the map. These effective trajectory nodes become the key positions for subsequent precise measurement and calculation.

[0114] Next, detect the acoustic wave parameters for these effective trajectory nodes. When acoustic waves propagate in media with different temperatures, parameters such as their propagation speed and frequency will change. Use high-precision acoustic wave detection equipment to measure parameters such as the propagation time and frequency of acoustic waves at the effective trajectory nodes. For example, through measurement, it is found that the acoustic wave propagation time at a certain effective trajectory node is 0.01 seconds (this data is obtained based on the time difference between the emission and reception recorded by the acoustic wave transceiver). According to the acoustic velocity-temperature linear relationship (this relationship is obtained by those skilled in the art through a large number of experiments and theoretical analyses. In a specific environment, such as an indoor environment at normal temperature and pressure, the acoustic velocity and temperature satisfy a specific linear formula, assumed to be v = 331.4 + 0.6T, where v is the acoustic velocity in m / s and T is the temperature in °C), substitute the measured acoustic wave parameters for matching to determine the target linear relationship.

[0115] Finally, calculate the temperature values of the effective trajectory nodes according to the determined target linear relationship. In the above example, the known acoustic wave propagation time is 0.01 seconds. Assuming the acoustic wave propagation distance is 3.4 meters (obtained by multiplying the propagation time by the acoustic velocity which is approximately 340 m / s in this environment), then the acoustic velocity v = 340 m / s. Substituting it into the acoustic velocity-temperature linear relationship v = 331.4 + 0.6T, after calculation (340 = 331.4 + 0.6T, by transposing we get 0.6T = 340 - 331.4 = 8.6, and then calculating T = 8.6 ÷ 0.6 ≈ 14.3 °C), the temperature value of this effective trajectory node can be measured to be approximately 14.3 °C.

[0116] Through the above steps, based on multi-source acoustic wave data, the temperature values of each effective trajectory node can be accurately measured in the signal trajectory space, providing key data support for the subsequent construction of a complete and accurate temperature field.

[0117] Furthermore, step A3000 in the method provided by the embodiments of the present application includes:

[0118] A3070: For the non-trajectory area, with the preset distributed spacing as a constraint, locate the distribution points of the non-trajectory area.

[0119] A3080: For the distribution points, calculate the temperature values of each distribution point in the K-nearest neighbor interpolation manner.

[0120] A3090: Construct a static temperature field according to the temperature values of each distribution point.

[0121] A3100: Among them, the weighted calculation based on the distance of the K nearest neighbor points of the distribution points is used as the calculation method.

[0122] In the embodiments of the present application, the non-trajectory area refers to the spatial range in the target area that is not covered by the signal transmission trajectory. The static temperature field is a temperature distribution model constructed by performing K-nearest neighbor interpolation on the non-trajectory area.

[0123] In one embodiment, during the real-time reconstruction of the dynamic temperature field based on multi-source acoustic wave data, when the signal trajectory space is determined through the previous measurement and analysis of the acoustic wave transceiver array, there will be a situation where some areas belong to the non-trajectory area. At this time, the preset distributed spacing is used as a key constraint to locate the distribution points of the non-trajectory area. Suppose in a monitoring scenario, the preset distributed spacing is set to 1 meter. Taking this spacing as a standard, the non-trajectory area is evenly divided into grids, and the intersection points of the grids are the distribution points. In this way, in a non-trajectory area with an area of 100 square meters, approximately 121 distribution points can be located (assuming the area is approximately square, and the number of grids is calculated according to the area).

[0124] Next, for these distribution points, the temperature values of each distribution point are calculated by means of K-nearest neighbor interpolation. The core principle of K-nearest neighbor interpolation is to estimate the temperature of a distribution point by using the temperature values of K nearest neighbor points around the distribution point. For example, set the value of K to 5. For each distribution point, by calculating its distance from all the surrounding known temperature points. (These known temperature points can be valid trajectory nodes in the signal trajectory space, and their temperatures have been obtained through the linear measurement in step A3060), select the 5 points with the closest distance as the nearest neighbor points. When calculating the distance, the Euclidean distance formula is usually used. For example, for a distribution point P(x, y) and a known temperature point Q(x1, y1), the Euclidean distance between them Suppose there is a distribution point P, and the 5 nearest neighbor points found through calculation are A, B, C, D, and E respectively, and their temperature values are T A 、T B 、T C 、T D 、T E respectively, and the corresponding distances are d A 、d B 、d C 、d D 、d E respectively. According to the distance-based weighted calculation method of the K nearest neighbor points of the distribution point, the temperature value T P of the distribution point P is calculated by the following formula: Through this weighted calculation method, the nearer the nearest neighbor point is to the distribution point, the greater its influence on the temperature value.

[0125] Finally, according to the calculated temperature values of each distribution point, a static temperature field is constructed. By integrating the temperature values of all distribution points according to their spatial positions in the non-trajectory area, the static temperature distribution of this area at a certain moment can be depicted. For example, by visually displaying the temperature values of the 121 distribution points calculated above on a two-dimensional plane and using different colors to represent different temperature ranges, the static temperature field of the non-trajectory area can be intuitively seen, with the areas with higher temperatures represented in red and the areas with lower temperatures represented in blue.

[0126] Through the above steps, the entire K-nearest neighbor interpolation process from locating the distribution points in the non-trajectory area to calculating the temperature values and then constructing the static temperature field is completed, providing an important data basis for the subsequent reconstruction and analysis of the dynamic temperature field.

[0127] Furthermore, step A3000 in the method provided by the embodiments of the present application includes:

[0128] A3110: For the static temperature field, perform feature-directed evolution decisions under multiple parallel branches to determine the branch prediction results.

[0129] A3120: Couple the branch prediction results and integrate them with the static temperature field based on the time series to determine the reconstructed temperature field.

[0130] In the embodiments of the present application, the feature-directed evolution decision is a decision-making process for the static temperature field, which conducts targeted analysis and prediction based on features in different dimensions. The branch prediction results are the prediction conclusions output by each parallel branch for the future state of the temperature field after executing the feature-directed evolution decision.

[0131] Optionally, after completing the previous steps and constructing the static temperature field, start executing the dynamic evolution prediction. Taking a large shopping mall as an example, for the static temperature field, execute the feature-directed evolution decision under multiple parallel branches. Suppose 5 parallel branches are set. Branch 1 focuses on the spatial diffusion characteristics of temperature. In the static temperature field of the shopping mall at a certain moment, it is found that the temperature in the clothing area is relatively high. By analyzing the diffusion speed of temperature in the past under similar spatial layouts and heat sources, it is found that every 10 minutes, the heat will diffuse towards the surrounding area at a speed of about 0.3 meters per minute. Based on this prediction, within the next 30 minutes, the temperature within a range of 2 - 3 meters around the clothing area will rise by 2 - 3 °C. Branch 2 focuses on studying the temperature change law in the time series. Collect the temperature data of the shopping mall at different times of each day in the past week, and combine it with the current static temperature field. It is found that from 12:00 noon to 2:00 pm every day, due to direct sunlight and dense personnel, the temperature in the atrium of the shopping mall is 5 - 7 °C higher than other times. The current static temperature field shows that it is 11:50 am at this time and the temperature in the atrium is at a relatively low level. Thus, it is predicted that within the next 1 - 2 hours, the temperature in the atrium will gradually rise by 5 - 7 °C. Branch 3 focuses on the temperature change near special heat sources. For example, the stoves in the food court of the shopping mall are the main heat sources. According to the stove power and past monitoring data, after the stove is turned on, the temperature within 1 meter around it will rise by 3 - 5 °C every 5 minutes. The current static temperature field shows that the stove of a certain food store has been turned on for 10 minutes. It is predicted that within the next 5 minutes, the temperature within 1 meter around this store will rise by another 3 - 5 °C. Branch 4 considers the influence of air flow on the temperature field. The shopping mall is equipped with a ventilation system. According to the position of the ventilation openings, the wind speed (assuming the wind speed is 1.2 m / s) and the wind direction, combined with the temperature distribution in each area of the static temperature field, it is predicted that after the cold air enters from the ventilation openings, it will cause the temperature in the surrounding area to drop by 4 - 6 °C within 20 minutes. Branch 5 focuses on the influence of different materials on temperature. The heat conduction performance of metal shelves and wooden decoration materials in the shopping mall is different. By comparing historical data, it is found that under direct sunlight, the surface temperature of metal shelves rises 2 - 3 times faster than that of wooden decoration. The current static temperature field shows that some metal shelves are in the direct sunlight area. It is predicted that within the next 15 minutes, the surface temperature of the metal shelves will be 5 - 7 °C higher than that of the surrounding wooden decoration materials.

[0132] Execute feature - oriented evolution decision for each of the above - mentioned branches to obtain their respective branch prediction results. Next, couple these branch prediction results. Those skilled in the art assign certain weights to each branch prediction result. For example, at the initial stage of a fire, the weight of branch 3 (temperature change near a special heat source) is set to 0.3; during normal operation periods, the weight of branch 4 (influence of air flow) is set to 0.25, etc. Then, perform weighted fusion on the branch prediction results according to the weights. For example, for a certain corner area of a shopping mall, branch 1 predicts a temperature increase of 2°C, branch 2 predicts an increase of 1°C, branch 3 predicts an increase of 3°C, branch 4 predicts a decrease of 1°C, and branch 5 predicts an increase of 1.5°C. After weighted calculation (assuming the weights of each branch are 0.2, 0.15, 0.3, 0.25, 0.1 in sequence), it is concluded that the temperature in this area will increase by 1.6°C in the next period of time.

[0133] Finally, integrate based on the time series and the static temperature field. Starting from the current moment, with a 5 - minute time interval, gradually update the temperature field according to the coupled prediction results. In the first 5 - minute period, slightly adjust the temperature of each area in the static temperature field according to the prediction results; in the second 5 - minute period, continue to adjust in combination with the temperature change in the previous 5 - minute period and the new prediction results. And so on. As time goes by, construct a dynamically changing reconstructed temperature field.

[0134] By gradually constructing a dynamically changing reconstructed temperature field, it can reflect the change situation of the temperature field in real - time, providing accurate temperature information support for fire monitoring, early warning, and personnel evacuation, etc.

[0135] In summary, the dynamic temperature field real - time reconstruction method based on multi - source acoustic wave data provided by the embodiments of the present application has the following technical effects:

[0136] In the present application, a measurement network is constructed by deploying an acoustic wave transceiver array in the target area, the signal trajectory space is determined by using acoustic wave co - frequency transmission and echo reception, a verification database is constructed through fire dynamics simulation, based on the linear relationship between sound speed and temperature, combined with linear measurement and K - nearest neighbor interpolation, a reconstruction unit is developed on a visualization simulation platform for real - time analysis and dynamic evolution prediction, the reconstructed temperature field is determined, and a connection with a fire monitoring system is established to achieve real - time reconstruction and monitoring of the dynamic temperature field, achieving the technical effect of non - contact, accurate, and real - time acquisition of internal temperature field information in a high - temperature and dynamically changing complex environment.

[0137] Embodiment 2, as Figure 2 shown, based on the same inventive concept as in the foregoing Embodiment 1, the embodiment of the present application provides a dynamic temperature field real - time reconstruction system based on multi - source acoustic wave data. The system includes:

[0138] Signal trajectory space determination module 1, which is used to deploy an acoustic wave transceiver array inside the target area, construct a multi-source acoustic wave measurement network, trigger the co-frequency transmission and echo reception of acoustic signals, and determine the signal trajectory space. Among them, each acoustic wave transceiver is marked with a spatial position code.

[0139] Platform data center storage module 2, which is used to connect to the visualization simulation platform, perform fire dynamics simulation under multiple fire conditions for the target area, construct a verification database and store it in the platform data center.

[0140] Reconstructed temperature field determination module 3, which is used to develop a reconstruction unit in the visualization simulation platform based on the linear relationship between sound speed and temperature constrained by acoustic signal parameters, assist the verification database, perform real-time analysis and dynamic evolution prediction based on linear measurement and K-nearest neighbor interpolation for the signal trajectory space, and determine the reconstructed temperature field.

[0141] Linear measurement execution module 4, which is used to perform linear measurement based on acoustic trajectory points and interpolation compensation for non-trajectory areas with a preset distributed spacing as a constraint.

[0142] Furthermore, the platform data center storage module 2 is used to perform the following steps:

[0143] Receive and analyze the echo signal, determine the signal transmission trajectory. Among them, each signal transmission trajectory is marked with a flight time; construct a three-dimensional space with the target area, perform spatial distribution on the signal transmission trajectory, and determine the signal trajectory space.

[0144] Furthermore, the platform data center storage module 2 is used to perform the following steps:

[0145] Determine the dynamic elements, where the dynamic elements at least include fire source, air flow, thermal radiation, and natural convection; based on the dynamic elements, perform fire scene clustering for the target area to determine multiple fire conditions; for the multiple fire conditions, perform fire dynamics simulation to determine the fire rules, where the dynamic evolution of distributed temperature is the simulation target; integrate the fire rules under each fire condition to construct the verification database.

[0146] Furthermore, the linear measurement execution module 4 is used to perform the following steps:

[0147] Establish connections among the acoustic wave transceiver array, the visualization simulation platform, and the fire monitoring system; among them, the acoustic wave transceiver array performs front-end detection, the reconstruction unit and the verification database developed in the visualization simulation platform perform simulation and reconstruction decisions, and the fire monitoring system performs response and alarm.

[0148] Further, the reconstructed temperature field determination module 3 is used to perform the following steps:

[0149] Introduce multi-head self-attention targets, where at least the global feature, local feature, and dynamic feature are targeted; according to the multi-head self-attention targets, construct multiple parallel branches, and perform parallel integration as the predicted reconstruction area; use the real-time reconstruction area and the predicted reconstruction area as the reconstruction unit, where each branch performs directional decision-making and dynamic evolution prediction, and couples the results output by the branches as the reconstruction result.

[0150] Further, the reconstructed temperature field determination module 3 is used to perform the following steps:

[0151] According to the real-time reconstruction area, identify the signal trajectory space, and use the preset distributed spacing to identify the effective trajectory nodes of the signal transmission trajectory; for the effective trajectory nodes, detect acoustic wave parameters for acoustic velocity-temperature linear relationship matching to determine the target linear relationship; according to the target linear relationship, calculate the temperature values of the effective trajectory nodes.

[0152] Further, the reconstructed temperature field determination module 3 is used to perform the following steps:

[0153] For the non-trajectory area, use the preset distributed spacing as a constraint to locate the distribution points of the non-trajectory area; for the distribution points, use the K-nearest neighbor interpolation method to calculate the temperature values of each distribution point; according to the temperature values of each distribution point, construct a static temperature field; where the weighted calculation based on the distance of the K nearest neighbors of the distribution points is used as the calculation method.

[0154] Further, the reconstructed temperature field determination module 3 is used to perform the following steps:

[0155] For the static temperature field, perform feature directional evolution decision-making under multiple parallel branches to determine the branch prediction results; couple the branch prediction results, and integrate them based on the time series and the static temperature field to determine the reconstructed temperature field.

[0156] The dynamic temperature field real-time reconstruction system based on multi-source acoustic wave data provided by the embodiments of the present invention can execute the dynamic temperature field real-time reconstruction method based on multi-source acoustic wave data provided by any embodiment of the present invention, and has the corresponding functional modules and beneficial effects for executing the method.

[0157] Although the present application makes various references to certain modules in the system according to embodiments of the present application, however, any number of different modules can be used and run on the user terminal and / or the server. The various units and modules included are only divided according to functional logic, but are not limited to the above division, as long as the corresponding functions can be achieved; in addition, the specific names of the functional units are only for the convenience of mutual distinction and do not limit the protection scope of the present invention.

[0158] The above specific embodiments do not constitute a limitation on the protection scope of the present application. Those skilled in the art should understand that various modifications, combinations and substitutions can be made according to design requirements and other factors. Any modifications, equivalent substitutions and improvements made within the spirit and principle of the present application shall be included within the protection scope of the present application. In some cases, the actions or steps recited in the present application can be executed in a different order from that in the embodiments and still achieve the desired results. Additionally, the processes depicted in the drawings do not necessarily require the specific order or sequential order shown to achieve the desired results. In certain embodiments, multitasking and parallel processing are also possible or may be advantageous.

Claims

1. A real-time reconstruction method for dynamic temperature field based on multi-source acoustic wave data, characterized in that, The method includes: Deploy an acoustic wave transceiver array inside the target area, construct a multi-source acoustic wave measurement network, trigger the co-frequency emission and echo reception of acoustic signals, and determine the signal trajectory space. Among them, each acoustic wave transceiver is marked with a spatial position code; Connect to the visual simulation platform, perform fire dynamics simulations under multiple fire conditions for the target area, construct a verification database and store it in the platform data center; Based on the linear relationship between sound speed and temperature constrained by acoustic signal parameters, develop a reconstruction unit in the visual simulation platform, assist the verification database, perform real-time analysis and dynamic evolution prediction based on linear measurement and K-nearest neighbor interpolation for the signal trajectory space, and determine the reconstructed temperature field; With a preset distributed spacing as a constraint, perform linear measurement based on acoustic trajectory points and interpolation compensation for non-trajectory areas.

2. The real-time reconstruction method of the dynamic temperature field based on multi-source acoustic wave data according to claim 1, wherein, Determine the signal trajectory space, including: Receive and analyze the echo signal, determine the signal transmission trajectory. Among them, each signal transmission trajectory is marked with a flight time; Construct a three-dimensional space with the target area, distribute the signal transmission trajectories in space, and determine the signal trajectory space.

3. The real-time reconstruction method of the dynamic temperature field based on multi-source acoustic wave data according to claim 1, characterized in that, Perform fire dynamics simulations under multiple fire conditions for the target area and construct a verification database, including: Determine the dynamic elements, where the dynamic elements at least include a fire source, air flow, thermal radiation, and natural convection; Cluster the fire scenarios for the target area according to the dynamic elements, and determine multiple fire conditions; For the multiple fire conditions, perform fire dynamics simulations to determine the fire rules, where the dynamic evolution of distributed temperature is the simulation target; Integrate the fire rules under each fire condition to construct the verification database.

4. The real-time reconstruction method of the dynamic temperature field based on multi-source acoustic wave data according to claim 1, wherein Establish connections between the acoustic wave transceiver array, the visual simulation platform, and the fire monitoring system; Among them, the acoustic wave transceiver array performs front-end detection, the reconstruction unit and the verification database developed in the visual simulation platform perform simulation reconstruction decisions, and the fire monitoring system performs response alarms.

5. The real-time reconstruction method of the dynamic temperature field based on multi-source acoustic wave data according to claim 1, wherein, Develop a reconstruction unit in the visual simulation platform, including: Introduce multi-head self-attention targets, where at least the global feature, local feature, and dynamic feature are used as targets; According to the multi-head self-attention targets, construct multiple parallel branches and integrate them in parallel as the prediction reconstruction area; Use the real-time reconstruction area and the prediction reconstruction area as the reconstruction unit. Among them, each branch performs directional decision-making and dynamic evolution prediction, and the results output by the branches are coupled as the reconstruction result.

6. The real-time reconstruction method of the dynamic temperature field based on multi-source acoustic wave data according to claim 5, characterized in that Perform linear measurement for the signal trajectory space, including: According to the real-time reconstruction area, identify the signal trajectory space, and mark the effective trajectory nodes of the signal transmission trajectory with the preset distributed spacing; For the effective trajectory nodes, detect acoustic parameters for sound speed-temperature linear relationship matching to determine the target linear relationship; According to the target linear relationship, calculate the temperature values of the effective trajectory nodes.

7. The real-time reconstruction method of dynamic temperature field based on multi-source acoustic wave data according to claim 6, characterized in that Perform K-nearest neighbor interpolation, including: For non-trajectory areas, with the preset distributed spacing as a constraint, locate the distribution points of the non-trajectory areas; For the distribution points, calculate the temperature values of each distribution point in the way of K-nearest neighbor interpolation; Construct a static temperature field according to the temperature values of each distribution point. Among them, the weighted distance of the K nearest neighbors of the distribution points is used as the calculation method.

8. The real-time reconstruction method of the dynamic temperature field based on multi-source acoustic wave data according to claim 7, characterized in that, Perform dynamic evolution prediction, including: For the static temperature field, perform feature-oriented evolution decisions under multiple parallel branches to determine the branch prediction results; Couple the branch prediction results and integrate them based on the time series and the static temperature field to determine the reconstructed temperature field.

9. A real-time reconstruction system for a dynamic temperature field based on multi-source acoustic wave data, characterized in that For implementing the real-time reconstruction method of the dynamic temperature field based on multi-source acoustic wave data according to any one of claims 1-8, the system includes: A signal trajectory space determination module, configured to deploy an acoustic wave transceiver array inside the target area, construct a multi-source acoustic wave measurement network, trigger the co-frequency emission and echo reception of acoustic signals, and determine the signal trajectory space, where each acoustic wave transceiver is marked with a spatial position code; A platform data center storage module, configured to connect to a visualization simulation platform, perform fire dynamics simulations under multiple fire conditions for the target area, construct a verification database and store it in the platform data center; A reconstructed temperature field determination module, configured to develop a reconstruction unit in the visualization simulation platform with the linear relationship between sound speed and temperature constrained by acoustic signal parameters, assist the verification database, perform real-time analysis based on linear measurement and K-nearest neighbor interpolation for the signal trajectory space, and perform dynamic evolution prediction to determine the reconstructed temperature field; A linear measurement execution module, configured to perform linear measurement based on acoustic trajectory points and interpolation compensation for non-trajectory areas with a preset distributed spacing as a constraint.

Images