An indoor fire source position recognition method and system based on temperature sensor data fusion and a precise disposal method
By calibrating the temperature sensor and fitting the error model, combined with Gaussian kernel two-dimensional interpolation technology, the location and intensity of the fire source can be directly obtained from the sensor data. This solves the problems of high difficulty and high cost in the construction of datasets in existing technologies, and achieves efficient and accurate fire source identification, supporting the scientific handling of fire rescue.
Patent Information
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- 713TH RES INST OF CHINA STATE SHIPBUILDING CORP LTD
- Filing Date
- 2026-01-15
- Publication Date
- 2026-06-09
AI Technical Summary
Existing fire monitoring technologies face challenges in data acquisition and application costs, resulting in insufficient accuracy in identifying fire source location and intensity, making it difficult to provide effective support for fire rescue.
By calibrating the temperature sensor and fitting the error model, the location and intensity of the fire source are directly obtained from the sensor data using Gaussian kernel two-dimensional interpolation technology, avoiding the data set construction process. The fire source intensity is represented by the heat release rate of the fire source, and the fire source is precisely dealt with through the sprinkler head.
It achieves efficient identification of fire source location and intensity, reduces algorithm deployment costs, provides accurate fire source information for fire rescue, and prevents items in non-fire source locations from being damaged by extinguishing agents.
Smart Images
Figure CN122174598A_ABST
Abstract
Description
Technical Field
[0001] This invention relates to a method, system, and precise handling method for identifying indoor fire sources based on temperature sensor data fusion, belonging to the field of fire source identification technology. Background Technology
[0002] With rapid socio-economic development and accelerated urbanization, fire has increasingly become a major threat to human life and property. Indoor fires, in particular, are characterized by their concealment, suddenness, and significant danger, easily causing severe casualties and property damage. The complex structure of indoor environments further exacerbates the concealment and suddenness of fires, posing enormous challenges to firefighters' emergency evacuation and rescue efforts.
[0003] Traditional fire monitoring systems typically detect physical indicators such as flame temperature or smoke concentration, triggering an alarm when these indicators exceed preset thresholds. However, this approach only outputs simple alarm signals and cannot provide specific location information or combustion intensity data for the fire source, thus failing to effectively support firefighters in developing scientifically sound firefighting plans.
[0004] In recent years, with the rapid development of artificial intelligence technology, flame recognition methods based on video surveillance have gradually emerged. These methods, through image processing and deep learning algorithms, can achieve preliminary identification of the location and intensity of fire sources. However, such algorithms still face many bottlenecks in practical applications, the most significant of which is their dependence on large labeled datasets. Constructing datasets requires significant resources and time, and the training and optimization processes of the algorithms are complex, resulting in high application costs. This, to some extent, limits their widespread promotion and practical application.
[0005] In summary, existing fire monitoring technologies are inadequate in terms of detection accuracy, information provision capabilities, and cost-effectiveness. There is an urgent need for a more efficient, accurate, and practical method for identifying and locating fire source intensity to better serve fire rescue operations.
[0006] For example, the Chinese invention patent authorization announcement text with authorization number CN112270122B discloses a method for inverting and evaluating fire source parameters in building fires. This method first selects a building space and fire source parameter types, and installs sensors within this space to send detection data at set time intervals. It then sets up different fire scenarios and uses a building fire smoke diffusion model to simulate each fire scenario, obtaining simulated detection values from the sensors at each sampling time corresponding to different fire source parameter values. These detection values are preprocessed to construct a dataset. Next, a deep learning sequence model is established and trained using the dataset, resulting in a fire source parameter inversion model. When a fire actually occurs, the actual detection values from the sensors at multiple times are acquired, preprocessed, and then input into the fire source parameter inversion model. This model outputs the probability distribution of the fire source parameter values. This approach relies on a large number of fire models to construct a dataset for model training. However, simulating fire scenarios using a fire aversion diffusion model is difficult, and the influencing factors of fire scenarios are complex and constantly changing, making it difficult to guarantee the accuracy of the simulation.
[0007] The Chinese utility model patent authorization announcement (CN202838579U) discloses a system for estimating the location and intensity of a fire source. This system includes multiple sets of estimation devices, each set located within a zone of the building structure. These devices include: multiple monitoring devices that detect environmental parameters to obtain fire monitoring information; a computing and storage device that records static and dynamic structural information, simulates and predicts fire information, and communicates with computing and storage devices in neighboring zones to locally estimate the location and intensity of the fire source between paired zones. The computing and storage devices in multiple zones form a distributed network to share fire monitoring information, static and dynamic structural information, and collaboratively process and calculate global estimates of the fire source location and intensity. However, the system relies heavily on a large dataset of fire models for training, resulting in challenges in dataset construction and high application costs.
[0008] The paper "Building Artificial-Intelligence Digital Fire (AID-Fire) system: A real-scale demonstration" discloses a real-time fire source information prediction model. This approach leverages the algorithmic advantages of convolutional neural networks and long short-term memory networks to construct a real-time fire source intensity prediction model. This model uses sequential data received from temperature sensors within a building to accurately and in real-time predict the location and intensity of fire sources in a fire scenario. Fire scenarios are simulated using fire dynamics simulation software to obtain real-time sequential data from temperature sensors, establishing a fire scenario database. Fire scenario data analysis is performed to train the real-time fire source intensity prediction model. Examples are used to verify the model's accuracy, timeliness, and robustness. This approach establishes a large numerical database containing 533 fire scenarios with different fire source sizes, locations, and numbers, and trains a convolutional long short-term memory neural network model. However, it also faces challenges in dataset construction and application costs.
[0009] The paper "Research on Flame Detection and Localization Technology Based on Deep Learning" discloses a deep learning-based intelligent fire source detection and localization method. It acquires fire source images through a video surveillance system and uses an improved YOLOX-nano algorithm to identify the fire source in the images. This process requires labeling the acquired flame images to create a flame dataset. Furthermore, it utilizes binocular vision theory to obtain the distance information of the fire source in three-dimensional coordinates, thus achieving fire source localization. However, this approach, besides requiring the labeling of flame images used for model training, also necessitates the use of a 3D camera such as a binocular camera to acquire flame images, still facing challenges in dataset construction and application costs. Summary of the Invention
[0010] The purpose of this invention is to provide a method, system, and precise handling method for indoor fire source location identification based on temperature sensor data fusion, in order to solve the problems of high difficulty in data set construction and high application cost in existing technologies.
[0011] To achieve the above objectives, the present invention includes:
[0012] The present invention provides a method for identifying indoor fire source locations based on temperature sensor data fusion, comprising the following steps:
[0013] 1) Interpolate the local temperatures at each temperature sensor location obtained from the temperature sensors installed on the roof to obtain two-dimensional temperature field data of the roof;
[0014] 2) After determining that a fire has occurred, obtain in real time the high-temperature coordinates of the coordinates where the temperature has reached the set temperature range below the highest temperature;
[0015] 3) Continuously calculate the real-time location coordinates of the fire source, and average the horizontal and vertical coordinates of the current high temperature coordinate point to obtain the current fire source calculation coordinates. The real-time location coordinates of the fire source obtained in the first calculation are the corresponding current fire source calculation coordinates.
[0016] When performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current calculation coordinates of the fire source and the previous real-time location coordinates of the fire source are averaged.
[0017] 4) The real-time location coordinates of the fire source obtained from the current calculation are used as the location of the fire source currently identified.
[0018] Furthermore, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the horizontal and vertical coordinates of the current fire source calculation are averaged with the horizontal and vertical coordinates of the previous fire source real-time location coordinates.
[0019] Further, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the horizontal and vertical coordinates of the corresponding current fire source calculation coordinates are averaged with the horizontal and vertical coordinates of the previous fire source real-time location coordinates. This is done by adding the product of the horizontal and vertical coordinates of the previous fire source real-time location coordinates and the previous number of times to the corresponding current fire source calculation coordinates, and then dividing by the current number of times to obtain the horizontal and vertical coordinates of the current fire source real-time location coordinates.
[0020] Further, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current fire source calculation coordinates and the previous real-time fire source location coordinates are averaged as follows: the horizontal and vertical coordinates of the corresponding current fire source calculation coordinates are respectively added to the horizontal and vertical coordinates of all previous real-time fire source location coordinates, and then divided by the current number of times to obtain the horizontal and vertical coordinates of the current fire source location coordinates.
[0021] Furthermore, prior to step 1), the temperature sensor is calibrated as follows:
[0022] a) Place the temperature sensor in the temperature calibration chamber, and then set the set temperature inside the temperature calibration chamber;
[0023] b) When the temperature calibration chamber reaches the set temperature, multiple temperature values are collected using a temperature sensor, and the average of the multiple temperature values is calculated; then the error value between the average value and the set temperature is calculated.
[0024] c) Change the set temperature and repeat step b) to obtain the error value at different set temperatures;
[0025] d) Using the mean of multiple temperature values collected by the temperature sensor at the set temperature as the independent variable and the corresponding error value as the dependent variable, fit and obtain the error model of the temperature sensor.
[0026] e) The measured temperature obtained by the temperature sensor is corrected using an error model and used as the local temperature at the corresponding location.
[0027] Further, in step e), the error model is used to correct the error by substituting the measured temperature into the error model to obtain the error prediction value, and then adding the measured temperature to obtain the local temperature at the corresponding location.
[0028] Furthermore, in step 1), the Gaussian kernel two-dimensional interpolation method is used for interpolation.
[0029] Furthermore, in step 2), if the local temperature exceeds a set temperature threshold a set number of times, it is determined that a fire has occurred.
[0030] Furthermore, it also includes the calculation of the ignition source intensity, using the ignition source heat release rate. Indicates the intensity of the ignition source:
[0031]
[0032] Where T0 represents normal temperature and H is the height of the roof.
[0033] This invention provides a precise indoor fire source control method based on temperature sensor data fusion, which controls the activation of fire sprinkler heads at the fire source location; the fire source location is identified using an indoor fire source location identification method including the following steps:
[0034] 1) Interpolate the local temperatures at each temperature sensor location obtained from the temperature sensors installed on the roof to obtain two-dimensional temperature field data of the roof;
[0035] 2) After determining that a fire has occurred, obtain in real time the high-temperature coordinates of the coordinates where the temperature has reached the set temperature range below the highest temperature;
[0036] 3) Continuously calculate the real-time location coordinates of the fire source, and average the horizontal and vertical coordinates of the current high temperature coordinate point to obtain the current fire source calculation coordinates. The real-time location coordinates of the fire source obtained in the first calculation are the corresponding current fire source calculation coordinates.
[0037] When performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current calculation coordinates of the fire source and the previous real-time location coordinates of the fire source are averaged.
[0038] 4) The real-time location coordinates of the fire source obtained from the current calculation are used as the location of the fire source currently identified.
[0039] Furthermore, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the horizontal and vertical coordinates of the current fire source calculation are averaged with the horizontal and vertical coordinates of the previous fire source real-time location coordinates.
[0040] Further, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the horizontal and vertical coordinates of the corresponding current fire source calculation coordinates are averaged with the horizontal and vertical coordinates of the previous fire source real-time location coordinates. This is done by adding the product of the horizontal and vertical coordinates of the previous fire source real-time location coordinates and the previous number of times to the corresponding current fire source calculation coordinates, and then dividing by the current number of times to obtain the horizontal and vertical coordinates of the current fire source real-time location coordinates.
[0041] Further, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current fire source calculation coordinates and the previous real-time fire source location coordinates are averaged as follows: the horizontal and vertical coordinates of the corresponding current fire source calculation coordinates are respectively added to the horizontal and vertical coordinates of all previous real-time fire source location coordinates, and then divided by the current number of times to obtain the horizontal and vertical coordinates of the current fire source location coordinates.
[0042] Furthermore, prior to step 1), the temperature sensor is calibrated as follows:
[0043] a) Place the temperature sensor in the temperature calibration chamber, and then set the set temperature inside the temperature calibration chamber;
[0044] b) When the temperature calibration chamber reaches the set temperature, multiple temperature values are collected using a temperature sensor, and the average of the multiple temperature values is calculated; then the error value between the average value and the set temperature is calculated.
[0045] c) Change the set temperature and repeat step b) to obtain the error value at different set temperatures;
[0046] d) Using the mean of multiple temperature values collected by the temperature sensor at the set temperature as the independent variable and the corresponding error value as the dependent variable, fit and obtain the error model of the temperature sensor.
[0047] e) The measured temperature obtained by the temperature sensor is corrected using an error model and used as the local temperature at the corresponding location.
[0048] Further, in step e), the error model is used to correct the error by substituting the measured temperature into the error model to obtain the error prediction value, and then adding the measured temperature to obtain the local temperature at the corresponding location.
[0049] Furthermore, in step 1), the Gaussian kernel two-dimensional interpolation method is used for interpolation.
[0050] Furthermore, in step 2), if the local temperature exceeds a set temperature threshold a set number of times, it is determined that a fire has occurred.
[0051] Furthermore, it also includes the calculation of the ignition source intensity, using the ignition source heat release rate. Indicates the intensity of the ignition source:
[0052]
[0053] Where T0 represents normal temperature and H is the height of the roof.
[0054] Further, the distance between the sprinkler head and the identified fire source location is calculated, and the distances are sorted in ascending order to obtain a distance sequence. The sprinkler heads corresponding to the first set number of values in the distance sequence are selected and activated.
[0055] This invention discloses an indoor fire source location identification system based on temperature sensor data fusion, comprising a memory, a processor, and a computer program stored in the memory. The processor executes the computer program to implement an indoor fire source location identification method including the following steps:
[0056] 1) Interpolate the local temperatures at each temperature sensor location obtained from the temperature sensors installed on the roof to obtain two-dimensional temperature field data of the roof;
[0057] 2) After determining that a fire has occurred, obtain in real time the high-temperature coordinates of the coordinates where the temperature has reached the set temperature range below the highest temperature;
[0058] 3) Continuously calculate the real-time location coordinates of the fire source, and average the horizontal and vertical coordinates of the current high temperature coordinate point to obtain the current fire source calculation coordinates. The real-time location coordinates of the fire source obtained in the first calculation are the corresponding current fire source calculation coordinates.
[0059] When performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current calculation coordinates of the fire source and the previous real-time location coordinates of the fire source are averaged.
[0060] 4) The real-time location coordinates of the fire source obtained from the current calculation are used as the location of the fire source currently identified.
[0061] Furthermore, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the horizontal and vertical coordinates of the current fire source calculation are averaged with the horizontal and vertical coordinates of the previous fire source real-time location coordinates.
[0062] Further, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the horizontal and vertical coordinates of the corresponding current fire source calculation coordinates are averaged with the horizontal and vertical coordinates of the previous fire source real-time location coordinates. This is done by adding the product of the horizontal and vertical coordinates of the previous fire source real-time location coordinates and the previous number of times to the corresponding current fire source calculation coordinates, and then dividing by the current number of times to obtain the horizontal and vertical coordinates of the current fire source real-time location coordinates.
[0063] Further, in step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current fire source calculation coordinates and the previous real-time fire source location coordinates are averaged as follows: the horizontal and vertical coordinates of the corresponding current fire source calculation coordinates are respectively added to the horizontal and vertical coordinates of all previous real-time fire source location coordinates, and then divided by the current number of times to obtain the horizontal and vertical coordinates of the current fire source location coordinates.
[0064] Furthermore, prior to step 1), the temperature sensor is calibrated as follows:
[0065] a) Place the temperature sensor in the temperature calibration chamber, and then set the set temperature inside the temperature calibration chamber;
[0066] b) When the temperature calibration chamber reaches the set temperature, multiple temperature values are collected using a temperature sensor, and the average of the multiple temperature values is calculated; then the error value between the average value and the set temperature is calculated.
[0067] c) Change the set temperature and repeat step b) to obtain the error value at different set temperatures;
[0068] d) Using the mean of multiple temperature values collected by the temperature sensor at the set temperature as the independent variable and the corresponding error value as the dependent variable, fit and obtain the error model of the temperature sensor.
[0069] e) The measured temperature obtained by the temperature sensor is corrected using an error model and used as the local temperature at the corresponding location.
[0070] Further, in step e), the error model is used to correct the error by substituting the measured temperature into the error model to obtain the error prediction value, and then adding the measured temperature to obtain the local temperature at the corresponding location.
[0071] Furthermore, in step 1), the Gaussian kernel two-dimensional interpolation method is used for interpolation.
[0072] Furthermore, in step 2), if the local temperature exceeds a set temperature threshold a set number of times, it is determined that a fire has occurred.
[0073] Furthermore, it also includes the calculation of the ignition source intensity, using the ignition source heat release rate. Indicates the intensity of the ignition source:
[0074]
[0075] Where T0 represents normal temperature and H is the height of the roof.
[0076] The beneficial effects of this invention are as follows:
[0077] This invention provides an indoor fire source location identification method, a precise handling method, and an indoor fire source location identification system. This method eliminates the need for building a dataset; it only requires processing the temperature data collected by sensors to obtain the fire source location. Furthermore, it can determine the fire source intensity, aiding firefighters in developing appropriate firefighting plans. The fire extinguishing method can also further specify the optimal location of sprinkler heads based on the fire source location information to prevent damage to items not located at the fire source from the extinguishing agent. This invention is characterized by its simplicity and low computational cost, resulting in low algorithm deployment costs. Attached Figure Description
[0078] Figure 1 This is a schematic flowchart of the indoor fire source location identification method of the present invention;
[0079] Figure 2(a) is a schematic diagram of the system architecture and indoor structure of the indoor fire source location identification system of the present invention;
[0080] Figure 2(b) is a schematic diagram of the horizontal layout of the sensors in the indoor fire source location identification method of the invention;
[0081] Figure 3 This is a schematic diagram of the top-level temperature interpolation of the indoor fire source location identification method of the present invention;
[0082] Figure 4 This is a schematic diagram of the indoor fire extinguishing method of the present invention. Detailed Implementation
[0083] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be described in detail below with reference to the accompanying drawings and embodiments.
[0084] Currently, machine learning or deep learning methods are mainly used to obtain the location and intensity of fire sources. These methods have the following two main drawbacks:
[0085] (1) In terms of dataset construction, compared with generating and simulating fire scenarios using algorithms, in order to obtain ideal fire source location and intensity model parameters, it is necessary to ignite different types and quantities of combustibles at multiple locations in the room and collect temperature sensor data to construct the dataset. This requires the collection of a large amount of data and incurs high costs;
[0086] (2) In terms of algorithm deployment, a hardware platform with strong computing power is required to run the network model, which results in high hardware costs.
[0087] This invention does not require the construction of a dataset; it only requires processing based on the sensor-corrected temperature to obtain the location and intensity of the fire source, and has the advantages of being simple to implement and easy to deploy.
[0088] In the following specific implementation, the fire source location refers to the coordinates of the fire source in the horizontal plane; the fire source intensity, i.e. the size of the fire source, is expressed by the heat release rate; the heat release rate is usually used to evaluate the combustion performance of materials and the hazard of fire, and is an important parameter for measuring how quickly a material releases heat in a fire.
[0089] Implementation method 1 for indoor fire source location identification:
[0090] like Figure 1 The flowchart shown illustrates a method for identifying the location of an indoor fire source. The method of the present invention includes the following steps.
[0091] (1) Temperature sensor calibration.
[0092] To obtain accurate fire scene temperature data, the temperature sensors need to undergo error calibration. First, all temperature sensors are placed in a high-precision temperature calibration chamber, and then the chamber temperature is set to T. k=1 =50℃, when the incubator reaches the preset temperature T k=1 At that time, the temperature sensor continuously collected N. num =10 temperature values, and then use formula (1) to obtain the sensor's temperature at T k=1 Average temperature collected at the temperature Then, using formula (2), the sensor's T value is obtained. k-1 Error value at temperature .
[0093] (1)
[0094] (2)
[0095] In the formula, k represents the set temperature of the incubator.
[0096] Following the method described above, the temperature error values of the sensor were obtained when the temperature chamber temperatures were 200℃, 400℃, 600℃, 800℃, and 1000℃, and compared with T1 (i.e., T...).k=1 The temperature error values at different times constitute the temperature error sequence of the sensor under different temperature environments. .
[0097] The mean of multiple temperature values collected by the temperature sensor at the set temperature of the incubator is taken as the independent variable, and the corresponding temperature error is taken as the dependent variable. The error curve of the sensor is obtained by the least squares fitting method as shown in equation (3).
[0098] (3)
[0099] Substitute the temperature value obtained by the temperature sensor at any temperature into x in formula (3) to calculate y, and then add it to the sensor temperature value x to obtain the corrected ambient temperature value.
[0100] Assume that the number of temperature sensors to be installed is N. sensor If the value is 16, then error correction operations need to be performed on 16 sensors.
[0101] (2) Temperature sensor layout.
[0102] Assuming the interior dimensions are Temperature sensors are placed on the roof, and a data acquisition card acquires the sensor data, which is then transmitted to an information processing center for processing and display. The indoor structure and the layout of the temperature sensors are shown in Figures 2(a) and 2(b). Typically, temperature sensors are not installed near the edges of the walls ( Therefore, in order to collect temperature values as close to the wall as possible, the distance between the edge sensors (i.e., the outermost temperature sensors) and the wall should not exceed 2m, and the sensors away from the wall should be evenly distributed (i.e., the temperature sensors outside the outermost edge should be evenly arranged in multiple rows and columns), as shown in Figure 2(b).
[0103] (3) Sensor temperature data acquisition.
[0104] The sensor collects indoor temperature data in real time. When the sensor corrects the temperature data to be greater than 60°C for 10 consecutive times, the information processing center will indicate an abnormal temperature. When the sensor corrects the temperature data to be greater than 100°C for 10 consecutive times, the information processing center will indicate a fire and issue an alarm.
[0105] (4) Two-dimensional temperature field.
[0106] A Gaussian kernel two-dimensional interpolation method is used to interpolate the data from the array sensor to a two-dimensional temperature field. In two-dimensional space, assuming point... and a set of data points and the corresponding data values Therefore, the two-dimensional interpolation formula using the Gaussian kernel can be expressed as:
[0107] (4)
[0108] In the formula, It is the distance between data points. These are the parameters of the Gaussian kernel. It is at point The interpolation results, This indicates the total number of data points.
[0109] Set the pixel size of the two-dimensional temperature field image to... That is, each pixel value represents Temperature value per square centimeter. Number of times N is used to determine the location of the maximum temperature. max This avoids misjudging the location of the maximum temperature due to fluctuations in sensor values, thereby improving the accuracy of two-dimensional temperature field interpolation. Initially, let the maximum temperature location be determined by the number of iterations N. max =0, when the sensor's maximum temperature is between 60 and 100℃, record the coordinates of the location of the maximum temperature value. Considering that the probability of a fire source appearing near a wall in a certain indoor scenario is low, N is only set when the distance between the location of the maximum temperature and the nearest wall is greater than 2m. max Increment by 1 until N. max Interpolation begins when the temperature sensor reading is >30 and reaches 100℃; otherwise, no interpolation is performed. There is no sensor data in the area between the outermost edge of the array sensor and the wall. Before interpolating the two-dimensional temperature field in this area, boundary conditions need to be manually set. Therefore, before interpolating the two-dimensional temperature field of an indoor scene, rows 1 and 1600 and columns 1 and 1400 of the two-dimensional image matrix are set to a fixed room temperature value of 30.
[0110] The top-level two-dimensional temperature field image is obtained by interpolating the real-time sensor data using equation (4), as shown below. Figure 3 As shown in the figure (the numbered dots in the figure indicate the location of temperature sensor s).
[0111] (5) Location information of the fire source.
[0112] When the maximum temperature T max When the temperature is above 100℃, the temperature value is located at Coordinates of all points within the interval (High temperature coordinate point), and then continuously calculate the real-time location coordinates of the fire source. First, use formula (5) to obtain the first calculated real-time location coordinates of the fire source. .
[0113] (5)
[0114] In the formula, n represents the total number of coordinates that meet the requirements of the interval, and N represents the number of times the coordinates of the fire source are calculated; formula (5) is also the formula for calculating the coordinates of the fire source.
[0115] When performing the second and subsequent fire source location calculations, the N fire source locations are averaged to obtain the real-time fire source location coordinates. This is achieved by averaging the horizontal and vertical coordinates of the current fire source calculation with the horizontal and vertical coordinates of the previous real-time fire source location.
[0116] As a specific example of calculating the location of the second and subsequent fire sources (an example of an average processing method), the location of the second and subsequent fire sources is calculated using formula (6).
[0117] (6)
[0118] in, The coordinates of the current secondary fire source are obtained from the current two-dimensional temperature field. Calculate the coordinates of the current fire source location. When N=1, , .
[0119] (6) Intensity of the fire source.
[0120] The heat release rate of the fire source is used to characterize the intensity of the fire source, as shown in formula (7).
[0121] (7)
[0122] In the formula, T0 represents normal temperature, and H is the building height.
[0123] In this embodiment, the sensor temperature error curve is obtained according to the least squares fitting algorithm, and the sensor temperature data is calibrated. Based on this, the location and intensity of the fire source can be obtained through further processing.
[0124] In the two-dimensional temperature field interpolation step, before the interpolation operation, it is determined whether the edge pixels of the two-dimensional image matrix need to be set to a fixed value based on the number of times the maximum temperature location Nmax is reached.
[0125] When calculating the fire source location, obtain the coordinates of all pixels that meet the preset temperature range, and calculate the average value of the fire source coordinates at this moment (i.e., the calculated fire source coordinates). For subsequent moments, calculate the average value of the current fire source coordinates (i.e., the calculated fire source coordinates) and the average value of the real-time fire source coordinates before this moment, and use it as the final real-time fire source coordinates for the corresponding moment.
[0126] Implementation method 2 for indoor fire source location identification:
[0127] The only difference between this embodiment and the embodiment 1 of the indoor fire source location identification method is that step (5) fire source location information.
[0128] Specifically, when calculating the location of the fire source for the second and subsequent times, the fire source locations are averaged from N times to obtain the real-time location coordinates of the fire source. One specific implementation (another implementation of averaging) is to calculate the location of the fire source for the second and subsequent times using formula (8).
[0129] (8)
[0130] in, The coordinates of the current secondary fire source are obtained from the current two-dimensional temperature field. Calculate the coordinates of the current fire source location. When N=1, , .
[0131] The other steps and other parts of step (5) in this embodiment are the same as those in embodiment 1 of the indoor fire source location identification method, and will not be repeated here.
[0132] Implementation method of indoor fire source location identification system:
[0133] Referring to Figure 2(a), the indoor fire source location identification system of the present invention serves as an information processing center in this embodiment. Specifically, it can be a computer or server, including a processor and a memory. The computer or server is also connected to a temperature sensor arranged on the roof via a data acquisition card. The processor executes a computer program stored in the memory to implement the steps of the indoor fire source location identification method of the present invention. The indoor fire source location identification method of the present invention has been described sufficiently clearly in the embodiments of the indoor fire source location identification method, and will not be repeated here.
[0134] Implementation of precise treatment methods:
[0135] Based on the implementation method of indoor fire source location identification, and referring to Figure 4 As shown, after completing step (5) fire source location, or after completing step (5) fire source location and step (6) fire source intensity identification, the following step (7) is also included.
[0136] (7) Spraying operation.
[0137] Assume the first The coordinates of the spray head positions are: Calculate the distance between all sprinkler heads and the fire source using formula (9). .
[0138] (9)
[0139] In the formula, , This indicates the total number of sprinkler heads.
[0140] Arranged in ascending order of numerical values Sort the distances to obtain a new distance sequence. Then select The first k (usually k=4) sprinkler heads are used to spray extinguishing agent, that is, the k sprinkler heads closest to the fire source carry out fire extinguishing operations to prevent items in non-fire source locations from being damaged by the extinguishing agent.
[0141] All other steps in this embodiment are the same as in Embodiment 1 of the Indoor Fire Source Location Identification Method, and will not be repeated here.
Claims
1. A method for identifying indoor fire source locations based on temperature sensor data fusion, characterized in that, Includes the following steps: 1) Interpolate the local temperatures at each temperature sensor location obtained from the temperature sensors installed on the roof to obtain two-dimensional temperature field data of the roof; 2) After determining that a fire has occurred, obtain in real time the high-temperature coordinates of the coordinates where the temperature has reached the set temperature range below the highest temperature; 3) Continuously calculate the real-time location coordinates of the fire source, and average the horizontal and vertical coordinates of the current high temperature coordinate point to obtain the current fire source calculation coordinates. The real-time location coordinates of the fire source obtained in the first calculation are the corresponding current fire source calculation coordinates. When performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current calculation coordinates of the fire source and the previous real-time location coordinates of the fire source are averaged. 4) The real-time location coordinates of the fire source obtained from the current calculation are used as the location of the fire source currently identified.
2. The indoor fire source location identification method based on temperature sensor data fusion according to claim 1, characterized in that, In step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the horizontal and vertical coordinates of the current fire source calculation coordinates are averaged with the horizontal and vertical coordinates of the previous fire source real-time location coordinates. The averaging process is as follows: add the product of the previous real-time location coordinates of the fire source and the previous number to the corresponding horizontal and vertical coordinates of the current fire source calculation coordinates, and then divide by the current number to obtain the horizontal and vertical coordinates of the current fire source real-time location coordinates.
3. The indoor fire source location identification method based on temperature sensor data fusion according to claim 1, characterized in that, In step 3), when performing the second and subsequent calculations of the real-time location coordinates of the fire source, the corresponding current fire source calculation coordinates and the previous real-time fire source location coordinates are averaged as follows: the horizontal and vertical coordinates of the corresponding current fire source calculation coordinates are respectively added to the horizontal and vertical coordinates of all previous real-time fire source location coordinates, and then divided by the current number of times to obtain the horizontal and vertical coordinates of the current fire source location coordinates.
4. The indoor fire source location identification method based on temperature sensor data fusion according to claim 1, characterized in that, Prior to step 1), the temperature sensor is calibrated as follows: a) Place the temperature sensor in the temperature calibration chamber, and then set the set temperature inside the temperature calibration chamber; b) When the temperature calibration chamber reaches the set temperature, multiple temperature values are collected using a temperature sensor, and the average of the multiple temperature values is calculated. Then calculate the error between the mean and the set temperature; c) Change the set temperature and repeat step b) to obtain the error value at different set temperatures; d) Using the mean of multiple temperature values collected by the temperature sensor at the set temperature as the independent variable and the corresponding error value as the dependent variable, fit and obtain the error model of the temperature sensor. e) The measured temperature obtained by the temperature sensor is corrected using an error model and used as the local temperature at the corresponding location.
5. The indoor fire source location identification method based on temperature sensor data fusion according to claim 4, characterized in that, In step e), the error model is used to correct the error by substituting the measured temperature into the error model to obtain the error prediction value, and then adding the measured temperature to obtain the local temperature at the corresponding location.
6. The indoor fire source location identification method based on temperature sensor data fusion according to claim 1, characterized in that, In step 1), interpolation is performed using a Gaussian kernel two-dimensional interpolation method; in step 2), a fire is determined to have occurred if the local temperature exceeds a set temperature threshold a set number of times.
7. The indoor fire source location identification method based on temperature sensor data fusion according to claim 1, characterized in that, It also includes the calculation of ignition source intensity, using the ignition source heat release rate. Indicates the intensity of the ignition source: Where T0 represents normal temperature and H is the height of the roof.
8. A method for precise handling of indoor fire sources based on temperature sensor data fusion, characterized in that, The fire sprinkler head at the location of the fire source is activated; the location of the fire source is identified using the indoor fire source location identification method as described in any one of claims 1 to 7.
9. The method for precise handling of indoor fire sources based on temperature sensor data fusion according to claim 8, characterized in that, Calculate the distance between the sprinkler head and the identified fire source location, sort the distances in ascending order to obtain a distance sequence, and select the sprinkler heads corresponding to the first set number of values from the distance sequence to start.
10. An indoor fire source location identification system based on temperature sensor data fusion, characterized in that, The device includes a memory, a processor, and a computer program stored in the memory, characterized in that the processor executes the computer program to implement the steps of the indoor fire source location identification method based on temperature sensor data fusion as described in any one of claims 1 to 7.