A building fire alarm method and system

Abstract

The application relates to the technical field of fire alarm, and particularly discloses a building fire alarm method and system, which comprises the following steps: step S1: determining a region needing to be monitored by acquiring shooting images and thermal imaging images of all monitoring points in a building and combining a target object library established in advance; step S2: collecting a smoldering sample image set of each target object and classifying the smoldering sample image set, and training a smoldering heat diffusion model of each class of target object by using a neural network model to simulate the heat diffusion characteristics; and step S3: dividing the monitoring region into pixel points, detecting whether a heat spot exists, acquiring initial thermal imaging of the heat spot if the heat spot is found, and continuously monitoring the heat spot, then calling a corresponding smoldering heat diffusion model to generate a comparison image set, and judging whether the heat spot is a smoldering source by comparing real-time thermal imaging with a model prediction result; the method realizes accurate detection of early smoldering sources of building fires by combining image recognition and a heat diffusion model.

Description

TECHNICAL FIELD

[0001] The present application relates to the field of fire alarm technology, in particular to a building fire alarm method and system. BACKGROUND

[0002] With the vigorous development of the construction industry, the structure of modern buildings is becoming more and more complex and diverse, and the internal space layout, decoration materials and various electrical equipment all show a high degree of diversification. In such an environment, fire hazards have become more hidden and elusive. As a fire form with strong concealment, the occurrence and development process of smoldering fire is more complex and difficult to detect than open fire.

[0003] The existing fire warning technology, although can monitor and warn fire to some extent, but when facing smoldering fire, it exposes many limitations. On the one hand, the traditional smoke detector mainly relies on the change of smoke concentration to trigger the alarm, and the smoldering fire often does not produce a lot of obvious smoke in the initial stage, which makes it difficult for the smoke detector to capture the fire signal in time. On the other hand, as one of the common fire warning devices, the temperature sensor works on the principle of judging the occurrence of fire based on the significant change of environmental temperature. However, smoldering fire usually only produces a relatively low temperature rise in the local area in the initial stage, and this temperature change may be relatively slow and easily covered by the normal temperature fluctuations of the surrounding environment, resulting in that the temperature sensor cannot accurately identify and timely issue an alarm. SUMMARY

[0004] The purpose of the present application is to provide a building fire alarm method and system, which solves the following technical problems.

[0005] The purpose of the present application can be achieved by the following technical solutions:

[0006] A building fire alarm method, comprising the following steps:

[0007] Step S1: acquiring all monitoring points in the building, acquiring a shooting image and a thermal imaging image at the monitoring points; establishing a target object library, the target object library containing a plurality of target objects; determining a monitoring area according to the target object library, the shooting image and the thermal imaging image;

[0008] Step S2: acquiring a smoldering sample image set of each target object, dividing the smoldering sample image set of each target object into a plurality of categories; establishing an initial model based on a neural network model, inputting the smoldering sample image set of all target objects in the same category into the initial model to obtain a smoldering heat diffusion model of each target object in the category;

[0009] Step S3: dividing the monitoring area into a plurality of pixel points, judging whether there is a hot spot in each pixel point, if there is a hot spot, obtaining an initial thermal imaging of the hot spot, and monitoring the hot spot in real time to obtain a thermal imaging set of the hot spot;

[0010] determining a smoldering heat diffusion model corresponding to the monitoring area, denoted as a contrast diffusion model; inputting the initial thermal imaging into the contrast diffusion model to obtain a contrast image set; comparing the thermal imaging set and the contrast image set to determine whether the hot spot is a smoldering source.

[0011] As a further scheme of the present application, the obtaining process of the photographed image and the thermal imaging image comprises:

[0012] A thermal imaging monitoring point is arranged at each monitoring point, the monitoring point is used to obtain a photographed image in a monitoring range, and the thermal imaging monitoring point is used to obtain a thermal imaging image of the monitoring range.

[0013] As a further scheme of the present application, the establishing process of the target object library comprises:

[0014] A sample image of each target object is obtained, image features of all sample images of the target object are extracted based on image recognition technology, denoted as sample image features, each sample image feature is associated with a corresponding target object, and a target object library is obtained.

[0015] As a further scheme of the present application, the determining process of the monitoring area comprises:

[0016] All objects in the photographed image are recognized, and an area occupied by the object in the photographed image is denoted as an object area; image features of the object area are extracted based on image feature extraction technology, denoted as object features; the object features are compared with all sample image features in the target object library, similarity degrees of the object features and each sample image feature are obtained, a plurality of sample image features with a similarity degree higher than a preset similarity threshold value are screened out, and a sample image feature with a highest similarity degree is selected from the plurality of sample image features, denoted as a target image feature; a target object corresponding to the target image feature in the target object library is obtained, and an area occupied by the object in the photographed image is denoted as a monitoring area.

[0017] As a further scheme of the present application, the obtaining process of the smoldering sample image set of the target object comprises:

[0018] Set a monitoring time period of a preset time length, record a time when the target object just starts to smolder as a smoldering start time, obtain a video stream of the target object in the monitoring time period after the smoldering start time from the smoldering start time of the target object; divide the video stream into a plurality of image frames at equal intervals to obtain a smoldering sample image set from all the image frames.

[0019] As a further scheme of the present application, the process of dividing the smoldering sample image set of each target object into a plurality of categories comprises:

[0020] Obtain the material of each target object, obtain the physical properties of each material, the physical properties including ignition point, specific heat capacity and porosity; preset the weight coefficient of each physical property, and obtain the attribute value of each material , wherein w i represents the weight coefficient of the i-th physical property, P i represents the value of the i-th physical property of the material, and n is the total number of physical properties.

[0021] Set a plurality of attribute value intervals, record a plurality of materials belonging to the same attribute value interval as a category, and record a plurality of target objects belonging to the same category as a category.

[0022] As a further scheme of the present application, the process of obtaining the smoldering heat diffusion model comprises:

[0023] Obtain the smoldering area of the target object in the image frame, obtain the temperature characteristics and geometric characteristics of the smoldering area, the temperature characteristics including the highest temperature, temperature gradient and temperature change rate of the smoldering area, and the geometric characteristics including the boundary change, area and diffusion speed of the smoldering area;

[0024] According to the monitoring time period, establish a time sequence, obtain the corresponding time nodes of each image frame on the time sequence; record the temperature characteristics and geometric characteristics at each time node as a training set, input the training set into the initial model, train the initial model, and obtain the smoldering heat diffusion model of the target object.

[0025] As a further scheme of the present application, a building fire alarm system comprises:

[0026] The data acquisition module comprises monitoring points and thermal imaging monitoring points, the thermal imaging monitoring points are arranged at the monitoring points, the monitoring points are used to obtain shooting images in a monitoring range, and the thermal imaging monitoring points are used to obtain thermal imaging images of the monitoring range.

[0027] Establish a target object library, the target object library contains a plurality of target objects; determine a monitoring area according to the target object library, shooting images and thermal imaging images;

[0028] The fitting module: obtains the smoldering sample image set of each target object, divides the smoldering sample image set of each target object into several categories; establishes an initial model based on a neural network model, inputs the smoldering sample image set of all target objects in the same category into the initial model, and obtains the smoldering heat diffusion model of each target object in the category;

[0029] The monitoring module: divides the monitoring area into several pixel points, judges whether there is a hot spot in each pixel point, if there is a hot spot, obtains the initial thermal image of the hot spot, and monitors the hot spot in real time to obtain a thermal image set of the hot spot;

[0030] The smoldering heat diffusion model corresponding to the monitoring area is determined, denoted as a comparative diffusion model; the initial thermal image is input into the comparative diffusion model to obtain a comparative image set; the thermal image set and the comparative image set are compared to determine whether the hot spot is a smoldering source.

[0031] The beneficial effects of the present application are:

[0032] The present application can effectively identify early smoldering sources by combining thermal imaging technology and neural network models, solving the problem that traditional smoke detectors and temperature sensors cannot detect smoldering fires in the early stage; the smoldering heat diffusion model is used for real-time comparative analysis of the hot spot, and the regression straight line slope is used to determine whether it is a smoldering source, significantly improving the accuracy and timeliness of fire warning; the model is trained according to the physical properties of the target object (such as ignition point, specific heat capacity, etc.), making the smoldering heat diffusion model more targeted and improving the adaptability in different scenarios; the monitoring area is automatically determined through image recognition, combined with pixel-level hot spot detection and region growing algorithm, reducing manual intervention and realizing automatic monitoring and early warning of fire hazards; the model is trained by combining temperature features and geometric features, enhancing the comprehensiveness and reliability of fire prediction; the present application significantly improves the early warning ability of building fires, especially smoldering fires, through technical innovation, and has important practical application value; and it is suitable for complex building environments, can accurately locate the position of fire hazards and provide real-time warning, providing efficient technical support for fire prevention and control. BRIEF DESCRIPTION OF DRAWINGS

[0033] The present application will be further described below with reference to the accompanying drawings.

[0034] Figure 1 is a schematic diagram of the steps of a building fire alarm method of the present application. DETAILED DESCRIPTION

[0035] With reference to the drawings of the embodiments of the present application, the technical solutions in the embodiments of the present application will be clearly and completely described, obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. Based on the embodiments of the present application, all other embodiments obtained by those skilled in the art without creative labor fall within the scope of protection of the present application.

[0036] Please refer to Figure 1 The present application is a kind of building fire alarm method, comprising the following steps:

[0037] Step S1: acquiring all monitoring points in the building, setting up thermal imaging monitoring points at each monitoring point, the monitoring point is used to acquire the shooting image in the monitoring range, and the thermal imaging monitoring point is used to acquire the thermal imaging image of the monitoring range;

[0038] A target object library is established, the target object library contains a plurality of target objects; based on image recognition technology and the target object library, all target objects in the shooting image are identified, and the area occupied by the target object in the thermal imaging image is marked, which is recorded as a monitoring area;

[0039] According to the building plan, monitoring points are deployed in key areas such as corridors, rooms and equipment rooms to ensure that the field of view of each monitoring point covers no dead angle; each monitoring point is integrated with a visible light camera (resolution ≥ 1080p) and an infrared thermal imager (temperature measurement range -20℃ to 550℃, accuracy ±2℃), and the timestamps of the two types of images are aligned through a synchronous controller; the visible light camera collects RGB images at a frame rate of 5fps, and the thermal imager synchronously outputs temperature matrix data, all data is preprocessed (denoising, distortion correction) through an edge computing node, and uploaded to a central server;

[0040] As a preferred embodiment of the present application, the target object is all objects in the building that have the risk of smoldering;

[0041] Collect multi-angle sample images of common flammable materials in the building, the target object includes wooden furniture, curtains, electrical equipment, etc., each type of object contains at least 200 labeled images, and the sample images are labeled with boundary boxes and material labels;

[0042] As a preferred embodiment of the present application, the establishment process of the target object library includes:

[0043] Obtain sample images of all target objects, extract image features of all sample images of the target objects based on image recognition technology, record the sample image features as sample image features, associate each sample image feature with its corresponding target object, and obtain a target object library;

[0044] Extract image features (such as SIFT, HOG, or deep features) using ResNet-50 or YOLOv7 models to build a feature vector database; compare real-time image features with library features using a cosine similarity algorithm, set a similarity threshold (such as 0.85), and select potential target objects; perform non-maximum suppression (NMS) on the matching results to eliminate duplicate detections;

[0045] As a preferred embodiment of the present application, the process of identifying the target object in the photographed image includes:

[0046] Identify all objects in the photographed image and record the area occupied by the object in the photographed image as the object area; extract image features of the object area based on image feature extraction technology and record them as object features; compare the object features with all sample image features in the target object library, obtain the similarity of the object features with each sample image feature, select a number of sample image features with a similarity to the object features exceeding a preset similarity threshold, and select the sample image feature with the highest similarity to the object features from the number of sample image features and record it as the target image feature; obtain the target object corresponding to the target image feature in the target object library, and record the area occupied by the object in the photographed image as the monitoring area;

[0047] Map the coordinates of the target object identified in the visible light image to the thermal imaging image to generate an ROI (region of interest) mask, perform adaptive threshold segmentation on the temperature data within the ROI to exclude environmental heat source interference, assign monitoring weights according to the fire risk level of the target object (such as material ignition point, historical fire data), use higher sampling frequency (10 fps) in high-risk areas, and reduce to 1 fps in low-risk areas to save computing power; the high-risk area includes a distribution box, and the low-risk area includes a metal support;

[0048] Apply HSV color space analysis in the visible light image to identify strong light areas or shadow areas, dynamically adjust the temperature difference sensitivity of the thermal imaging, and if the temperature fluctuation in a certain area is <±1℃ within 30 seconds, mark it as a steady-state environment and suspend real-time monitoring;

[0049] Step S2: Obtain a set of smoldering sample images for each target object, divide the set of smoldering sample images for each target object into several categories; based on a neural network model, an initial model is established, the set of smoldering sample images for all target objects in the same category is input into the initial model, and after training the initial model, the smoldering heat diffusion model for each target object in the category is obtained;

[0050] As a preferred embodiment of the present application, the process of obtaining the set of smoldering sample images for the target object includes:

[0051] Set a preset time length of a monitoring time period, mark a time when the target object just starts to smolder as a smoldering start time, obtain a video stream of the target object in the monitoring time period after the smoldering start time from the smoldering start time of the target object; the video stream is equally spaced into a plurality of image frames, and a smoldering sample image set is obtained from all the image frames;

[0052] It should be noted that a 1:1 building material test platform is built in a controlled laboratory environment, a constant temperature and humidity system is configured, the temperature range in the controlled laboratory environment is 23±2 DEG C, and the humidity range is 50±5% RH; an industrial thermal imager is used, and data is collected synchronously with a 4K high-speed camera, 3 sets of control experiments are set for each target material, and are respectively normal, smoldering initial stage and open fire conversion stage;

[0053] As a preferred embodiment of the present application, the process of dividing the smoldering sample image set of each target object into a plurality of categories includes:

[0054] Obtain the material of each target object, obtain the physical properties of each material, the physical properties include ignition point, specific heat capacity and porosity; preset the weight coefficient of each physical property, and obtain the attribute value of each material , wherein w i represents the weight coefficient of the i-th physical property, P i represents the numerical value of the i-th physical property of the material, and n is the total number of physical properties;

[0055] A plurality of attribute value intervals are set, and a plurality of materials with the same attribute value interval are recorded as a category; and a plurality of target objects with the same category are recorded as the same category.

[0056] The ignition point determines the smoldering start temperature threshold, the specific heat capacity reflects the heat accumulation speed, and the porosity affects the oxygen diffusion efficiency.

[0057] The attribute value of the material fire risk is calculated by using a linear weighting formula, and the weight coefficient is determined by using an analytic hierarchy process or principal component analysis; the Av value is divided into 3 to 5 intervals by using K-means clustering or equal frequency binning method.

[0058] As a preferred embodiment of the present application, the process of obtaining the smoldering heat diffusion model includes:

[0059] Obtain the smoldering area of the target object in the image frame, obtain the temperature characteristics and geometric characteristics of the smoldering area, the temperature characteristics include the highest temperature, temperature gradient and temperature change rate of the smoldering area, and the geometric characteristics include the boundary change, area and diffusion speed of the smoldering area.

[0060] According to the monitoring time period, a time sequence is established, time nodes corresponding to each image frame on the time sequence are obtained, temperature features and geometric features at each time node are recorded as a training set, the training set is input into the initial model, the initial model is trained, and a smoldering heat diffusion model of the target object is obtained;

[0061] It should be noted that, for different materials, a dedicated smoldering heat diffusion model is generated for quantitative prediction; by synchronously analyzing the thermal imaging temperature field, visible light shape change and material physical property parameters, a complete feature expression of the smoldering process is constructed; the spatiotemporal evolution law of temperature diffusion is captured by using a 3D convolutional neural network, the spatiotemporal evolution law includes heat flow direction and speed, and the combustion trend is predicted in combination with material physical properties; the heat transfer equation is embedded into the neural network loss function, the model is constrained to comply with the physical law, and the predicted temperature change rate is ensured to be consistent with the actual thermodynamic process through diffusion consistency loss;

[0062] Step S3: dividing the monitoring area into a plurality of pixel points, judging whether there is a hot spot in each pixel point, if there is a hot spot, obtaining an initial thermal image of the hot spot, and monitoring the hot spot in real time to obtain a thermal image set of the hot spot;

[0063] A smoldering heat diffusion model corresponding to the monitoring area is determined, denoted as a comparative diffusion model; the initial thermal image is input into the comparative diffusion model to obtain a comparative image set; the thermal image set and the comparative image set are compared to determine whether the hot spot is a smoldering source;

[0064] The monitoring area is divided into high-resolution grids, the temperature of each pixel point is collected in real time by a thermal imager, an adaptive threshold algorithm is used to identify an abnormally high temperature area, i.e., a hot spot; time sequence tracking is performed on the candidate hot spot, which needs to exist continuously for 5 frames (about 1 second) and expand in area to be confirmed as an effective hot spot;

[0065] As a preferred embodiment of the present application, the process of judging whether there is a hot spot in each pixel point includes:

[0066] The average temperature of each pixel point in the smoldering area at the smoldering starting time of the target object is obtained, the minimum value of the average temperature of each target object is obtained, and a temperature threshold is set according to the minimum value; if the temperature of a pixel point in the monitoring area exceeds the temperature threshold, the pixel point is recorded as a hot spot;

[0067] It should be noted that if the hot spot can be composed of multiple adjacent pixels, a region growing algorithm is used to merge the connected regions higher than the threshold into one hot spot;

[0068] The process of setting a temperature threshold according to the minimum value includes:

[0069] In the normal operation state of the building, temperature data of each monitoring point in the monitoring time period is continuously collected, and the smoldering starting temperature characteristics of various materials are obtained through laboratory simulation; a sliding window percentile method is used to calculate a real-time threshold value, and the temperature threshold value T th =T min +k× σ , wherein T min is a minimum value, σ is a temperature standard deviation of the monitoring point in the past 24 hours, used to eliminate the influence of environmental fluctuations, and k is a sensitivity coefficient, and the default is 2.5, and the high-risk area can be adjusted to 1.8;

[0070] As a preferred embodiment of the present application, the initial thermal image of the hot spot is the thermal image when the hot spot is first discovered;

[0071] As a preferred embodiment of the present application, the process of obtaining the thermal image set of the hot spot includes:

[0072] The thermal imaging monitoring point obtains the thermal imaging image of the monitoring area in real time, obtains a plurality of real-time thermal imaging images, and records all real-time thermal imaging images as a thermal image set;

[0073] As a preferred embodiment of the present application, the process of obtaining the contrast image set includes:

[0074] The contrast diffusion model takes the initial thermal image as the thermal image corresponding to the monitoring area at the smoldering starting moment; the time stamp corresponding to the initial thermal image is the initial moment, and the contrast diffusion model generates images of each time node in the monitoring time period after the initial moment for the monitoring area, recorded as simulation images, and a contrast image set is obtained from all simulation images;

[0075] As a preferred embodiment of the present application, the process of determining whether the hot spot is a smoldering source includes:

[0076] Each real-time thermal imaging image in the thermal image set is numbered, and each simulation image in the contrast image set is numbered; the similarity between each thermal imaging image and simulation image with the same number is obtained, recorded as image similarity;

[0077] A contrast similarity threshold value is set, a similarity difference value corresponding to each number is obtained in real time, the similarity difference value C=Sm-Sm´, wherein Sm is the image similarity, and Sm´ is the contrast similarity threshold value; the similarity difference value corresponding to each number is obtained, and based on the least square method, each number and the similarity difference value corresponding thereto are fitted to obtain a regression straight line; the slope of the regression straight line is obtained, if the slope is greater than 0, the hot spot is a smoldering source; otherwise, the hot spot is not a smoldering source;

[0078] The actual measurement thermal imaging and the model prediction are compared frame by frame using a structural similarity index (SSIM) or a Hausdorff distance, if the similarity is greater than 0.8 and the error gradient is consistent, it is determined as a smoldering source, if the similarity is less than 0.8 and the diffusion direction is inconsistent, it is determined as an interference source;

[0079] The number is used to eliminate the time offset of the model prediction and the measured data by dynamic time warping;

[0080] It can be understood that when the heat spot is a smoldering source, the position coordinates of the monitoring area in the building are acquired, and a warning is sent to the relevant staff or platform, prompting the relevant staff or platform that there is a smoldering fire hazard at the position coordinates.

[0081] A building fire alarm system comprises:

[0082] The data acquisition module comprises monitoring points and thermal imaging monitoring points, the thermal imaging monitoring points are arranged at the monitoring points, the monitoring points are used to acquire shooting images in a monitoring range, and the thermal imaging monitoring points are used to acquire thermal imaging images of the monitoring range;

[0083] A target object library is established, the target object library comprises a plurality of target objects; a monitoring area is determined according to the target object library, the shooting images and the thermal imaging images;

[0084] The fitting module acquires a smoldering sample image set of each target object, divides the smoldering sample image set of each target object into a plurality of categories, establishes an initial model based on a neural network model, inputs the smoldering sample image set of all target objects in the same category into the initial model, and obtains a smoldering heat diffusion model of each target object in the category;

[0085] The monitoring module divides the monitoring area into a plurality of pixel points, judges whether there is a heat spot in each pixel point, if there is a heat spot, an initial thermal image of the heat spot is acquired, and the heat spot is monitored in real time to obtain a thermal image set of the heat spot;

[0086] A smoldering heat diffusion model corresponding to the monitoring area is determined, denoted as a comparative diffusion model; the initial thermal image is input into the comparative diffusion model to obtain a comparative image set; the thermal image set and the comparative image set are compared to determine whether the heat spot is a smoldering source.

[0087] The above describes one embodiment of the present application in detail, but the content is the preferred embodiment of the present application, and cannot be considered as limiting the scope of the present application. Any equivalent changes and improvements made according to the scope of the present application should still belong to the patent scope of the present application.

Claims

1. A building fire alarm method, characterized in that, Includes the following steps: Step S1: Obtain all monitoring points within the building, and acquire captured images and thermal images at the monitoring points; Establish a target object database containing several target objects; determine the monitoring area based on the target object database, captured images, and thermal images; Step S2: Obtain the smoldering sample image set of each target object, and divide the smoldering sample image set of each target object into several categories; establish an initial model based on the neural network model, and input the smoldering sample image set of all target objects in the same category into the initial model to obtain the smoldering heat diffusion model of each target object in the category. Step S3: Divide the monitoring area into several pixels, determine whether there is a hot spot in each pixel, if there is a hot spot, obtain the initial thermal image of the hot spot, and monitor the hot spot in real time to obtain the thermal image set of the hot spot. A smoldering heat diffusion model corresponding to the monitoring area is determined and denoted as the contrast diffusion model; the initial thermal image is input into the contrast diffusion model to obtain a contrast image set; the thermal image set is compared with the contrast image set to determine whether the hot spot is a smoldering source.

2. The building fire alarm method according to claim 1, characterized in that, In step S1, the process of obtaining the captured image and the thermal image includes: Thermal imaging monitoring points are set up at each monitoring point. The monitoring points are used to acquire images captured within the monitoring range, and the thermal imaging monitoring points are used to acquire thermal images of the monitoring range.

3. The building fire alarm method according to claim 1, characterized in that, In step S1, the process of establishing the target object library includes: Obtain sample images of all target objects, extract image features of all sample images of the target objects based on image recognition technology, and denot them as sample image features. Associate each sample image feature with its corresponding target object to obtain a target object library.

4. A building fire alarm method according to claim 1, characterized in that, In step S1, the process of determining the monitoring area includes: All objects in the captured image are identified, and the area occupied by each object in the captured image is denoted as the object region. Image features of the object region are extracted based on image feature extraction technology and denoted as object features. The object features are compared with all sample image features in the target object database to obtain the similarity between the object features and each sample image feature. Several sample image features with a similarity exceeding a preset similarity threshold are selected, and the sample image feature with the highest similarity to the object feature is selected from these sample image features and denoted as the target image feature. The target object corresponding to the target image feature in the target object database is obtained, and the area occupied by the object in the captured image is denoted as the monitoring area.

5. A building fire alarm method according to claim 1, characterized in that, In step S2, the process of acquiring the smoldering sample image set of the target object includes: A preset monitoring period is set, and the moment when the target object just begins to smolder is recorded as the smoldering start time. Starting from the smoldering start time of the target object, the video stream of the target object within the monitoring period after the smoldering start time is acquired. The video stream is divided into several image frames at equal intervals, and a smoldering sample image set is obtained from all image frames.

6. A building fire alarm method according to claim 1, characterized in that, In step S2, the process of dividing the smoldering sample image set of each target object into several classes includes: The material of each target object is obtained, and the physical properties of each material are obtained, including ignition point, specific heat capacity, and porosity; weighting coefficients for each physical property are preset, and the property values ​​of each material are obtained. , where w i P represents the weight coefficient of the i-th physical attribute. i This represents the value of the i-th physical property of the material, where n is the total number of physical properties; Several attribute value ranges are defined, and materials whose attribute values ​​belong to the same attribute value range are categorized into one class; and several target objects whose materials belong to the same class are also categorized into the same class.

7. A building fire alarm method according to claim 5, characterized in that, In step S2, the process of obtaining the smoldering heat diffusion model includes: Within the image frame, the smoldering region of the target object is obtained, and the temperature and geometric features of the smoldering region are obtained. The temperature features include the highest temperature, temperature gradient, and temperature change rate of the smoldering region, and the geometric features include the boundary changes, area, and diffusion rate of the smoldering region. Based on the monitoring period, a time series is established, and the time nodes corresponding to each image frame in the time series are obtained; the temperature features and geometric features at each time node are recorded as the training set, and the training set is input into the initial model to train the initial model and obtain the smoldering heat diffusion model of the target object.

8. A building fire alarm method according to claim 2, characterized in that, In step S3, the process of determining whether the hot spot is a smoldering source includes: The thermal imaging monitoring point acquires thermal images of the monitoring area in real time, resulting in several real-time thermal images. All real-time thermal images are recorded as a thermal image set. The comparative diffusion model uses the initial thermal image as the thermal image of the monitored area at the moment of smoldering initiation; the timestamp corresponding to the initial thermal image is taken as the initial time, and the comparative diffusion model generates images of the monitored area at each time node during the monitoring period after the initial time, which are recorded as simulated images. A set of comparative images is obtained from all simulated images. Each real-time thermal image in the thermal image set is numbered, and each simulated image in the comparison image set is numbered; the similarity between each thermal image and simulated image with the same number is obtained and recorded as image similarity. A comparison similarity threshold is set, and the similarity difference corresponding to each number is obtained in real time. The similarity difference C = Sm - Sm', where Sm is the image similarity and Sm' is the comparison similarity threshold. The similarity difference corresponding to each number is obtained, and the least squares method is used to fit each number and its corresponding similarity difference to obtain a regression line. The slope of the regression line is obtained. If the slope is greater than 0, the hot spot is a smoldering source; otherwise, the hot spot is not a smoldering source.

9. A building fire alarm system, characterized in that, include: Data acquisition module: includes monitoring points and thermal imaging monitoring points. Thermal imaging monitoring points are set at each monitoring point. The monitoring points are used to acquire images within the monitoring range, and the thermal imaging monitoring points are used to acquire thermal images of the monitoring range. Establish a target object database containing several target objects; determine the monitoring area based on the target object database, captured images, and thermal images; Fitting module: acquires the smoldering sample image set of each target object, divides the smoldering sample image set of each target object into several classes; establishes an initial model based on a neural network model, inputs the smoldering sample image set of all target objects in the same class into the initial model, and obtains the smoldering heat diffusion model of each target object in the class. Monitoring module: Divide the monitoring area into several pixels, determine whether there are hot spots in each pixel, if there are hot spots, acquire the initial thermal image of the hot spots, and monitor the hot spots in real time to obtain the thermal image set of the hot spots; A smoldering heat diffusion model corresponding to the monitoring area is determined and denoted as the contrast diffusion model; the initial thermal image is input into the contrast diffusion model to obtain a contrast image set; the thermal image set is compared with the contrast image set to determine whether the hot spot is a smoldering source.

Images