A fire warning and automatic alarm method and system based on smart lamp poles

Abstract

The present invention relates to a fire warning and automatic alarm method and system based on a smart light pole. The method specifically includes: establishing a combustible material type and smoke refractive index template library, determining the combustible material type based on the smoke image captured by the camera of the smart light pole, obtaining a corresponding second standard smoke refractive index range based on the determined combustible material type, and optimizing the environmental parameter and smoke refractive index mapping model based on the second standard smoke refractive index range; identifying the smoke image captured by the camera of the smart light pole based on the environmental parameter and smoke refractive index mapping model to determine the smoke refractive index change value in the current environment; correcting the smoke refractive index change value in the current environment based on the smoke refractive index calibration model to obtain a corrected smoke refractive index change value; determining a fire probability value based on the corrected smoke refractive index change value, and judging whether a fire hazard exists based on the fire probability value. The present invention achieves accurate identification and rapid response to fire hazards.

Description

Technical Field

[0001] The present invention relates to the field of artificial intelligence technology, and in particular to a fire warning and automatic alarm method and system based on a smart lamp pole. Background Art

[0002] As smart city development continues to deepen, the intelligent and efficient transformation of fire prevention and early warning systems has become a critical issue that needs to be addressed. This transformation not only affects the safety of life and property of urban residents but also serves as a key cornerstone for building a safe, smart, and livable urban environment.

[0003] The core of traditional fire alarm systems lies in the deployment and application of temperature and smoke detectors. These hardware devices monitor temperature changes and smoke concentrations in the environment, triggering alarms once preset thresholds are reached. However, this model has exposed numerous limitations in practical applications. First, its response speed is often limited by the sensitivity of the detectors and the efficiency of data transmission. This makes it difficult to issue early warnings in the early stages of a fire, thus missing the optimal opportunity to extinguish the fire.

[0004] Another major challenge facing traditional fire alarm systems is the high false alarm rate. In complex and changing environmental conditions, such as smog-shrouded cities, strong winds in outdoor areas, and high-humidity indoor spaces, naturally occurring smoke, water vapor, and dust particles can easily trigger false alarms in detectors. This not only disrupts normal social order but also undermines public trust in fire alarm systems.

[0005] Furthermore, traditional fire alarm systems have a relatively limited coverage area. Limited by the location and number of detectors, systems often struggle to achieve comprehensive, comprehensive monitoring of every corner of a city. This is especially true in high-risk areas like high-rise buildings, underground spaces, and densely populated commercial areas, where blind spots in traditional equipment can become hotbeds for fire spread.

[0006] Furthermore, with the acceleration of urbanization and the impact of climate change, fire risks are becoming increasingly complex and dynamic. The interplay of different combustible materials, varying combustion conditions, and environmental factors makes the characteristics of fire smoke difficult to predict. Traditional fire alarm systems, lacking the ability to deeply analyze and identify fire smoke characteristics, often struggle to accurately distinguish between natural smoke and fire smoke in complex environments, leading to false alarms or missed alerts. Summary of the Invention

[0007] The object of the present invention is to provide a fire warning and automatic alarm method and system based on a smart lamp pole to solve at least one of the above-mentioned problems in the prior art.

[0008] In a first aspect, the present invention provides a fire warning and automatic alarm method based on a smart lamp pole, the method specifically comprising:

[0009] Based on the differences in the refractive index of smoke in different environments, a mapping model between environmental parameters and smoke refractive index is established, wherein the mapping model between environmental parameters and smoke refractive index is used to determine the first smoke refractive index range corresponding to different environments;

[0010] Establish a template library of combustible material types and smoke refractive indexes, determine the combustible material type based on the smoke image captured by the smart lamp pole camera, obtain the corresponding second standard smoke refractive index range based on the determined combustible material type, and optimize the environmental parameter and smoke refractive index mapping model based on the second standard smoke refractive index range;

[0011] Identify the smoke image captured by the camera of the smart lamp pole according to the environmental parameters and the smoke refractive index mapping model, and determine the change value of the smoke refractive index in the current environment;

[0012] Establishing a smoke refractive index calibration model based on the smoke refractive index under different lighting conditions, and correcting the smoke refractive index change value under the current environment according to the smoke refractive index calibration model to obtain a corrected smoke refractive index change value;

[0013] A fire probability value is determined based on the modified smoke refractive index change value, and whether there is a fire hazard is determined based on the fire probability value.

[0014] In a second aspect, the present invention provides a fire warning and automatic alarm system based on a smart lamp pole, the system specifically comprising:

[0015] A first processing module is configured to establish an environmental parameter and smoke refractive index mapping model based on the difference in smoke refractive index changes in different environments, wherein the environmental parameter and smoke refractive index mapping model is used to determine a first smoke refractive index range corresponding to different environments;

[0016] A second processing module is configured to establish a template library of combustible material types and smoke refractive indexes, determine the combustible material type based on the smoke image captured by the camera of the smart lamp pole, obtain a corresponding second standard smoke refractive index range based on the determined combustible material type, and optimize the environmental parameter and smoke refractive index mapping model based on the second standard smoke refractive index range;

[0017] a third processing module, configured to identify the smoke image captured by the camera of the smart lamp pole according to the environmental parameters and the smoke refractive index mapping model, and determine a change value of the smoke refractive index in the current environment;

[0018] a fourth processing module, configured to establish a smoke refractive index calibration model based on the smoke refractive index under different lighting conditions, and perform correction processing on the smoke refractive index change value under the current environment based on the smoke refractive index calibration model to obtain a corrected smoke refractive index change value;

[0019] The fifth processing module is used to determine a fire probability value according to the modified smoke refractive index change value, and determine whether there is a fire hazard according to the fire probability value.

[0020] In a third aspect, the present invention provides a computer device comprising: a memory and a processor and a computer program stored in the memory. When the computer program is executed on the processor, it implements a fire warning and automatic alarm method based on a smart lamp pole as described in any one of the above methods.

[0021] In a fourth aspect, the present invention provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the fire warning and automatic alarm method based on a smart lamp pole as described in any one of the above methods is implemented.

[0022] Compared with the prior art, the present invention has at least one of the following technical effects:

[0023] 1. By comprehensively applying advanced technical means such as environmental parameters and smoke refractive index mapping model, combustible material type and smoke refractive index template library, smoke refractive index calibration model and logistic regression model, accurate identification and rapid response to fire hazards are achieved.

[0024] 2. Based on the differences in smoke refractive index changes in different environments, a mapping model between environmental parameters and smoke refractive index was established. This model can accurately determine the smoke refractive index range in the current environment, thereby improving the accuracy of smoke recognition.

[0025] 3. By establishing a template library of combustible material types and smoke refractive index, and determining the combustible material types based on the smoke images captured by the camera, the environmental parameters and smoke refractive index mapping model are further optimized to make it more in line with actual conditions.

[0026] 4. The influence of lighting conditions on smoke recognition is taken into consideration. By establishing a smoke refractive index calibration model and correcting the smoke refractive index according to the light intensity in the current environment, the influence of changes in lighting conditions on smoke recognition is effectively eliminated.

[0027] 5. The fire probability value of the smoke image of preset continuous frames is calculated through the logistic regression model, and the presence of fire hazards is judged based on the preset fire probability threshold, realizing automatic identification and rapid alarm of fire hazards. BRIEF DESCRIPTION OF THE DRAWINGS

[0028] In order to more clearly illustrate the technical solutions in the embodiments of the present application, the following briefly introduces the drawings required for use in the embodiments. Obviously, the drawings described below are only some embodiments of the present application. For ordinary technicians in this field, other drawings can be obtained based on these drawings without any creative work.

[0029] Figure 1 This is a flow chart of a fire warning and automatic alarm method based on a smart lamp pole provided by one embodiment of the present invention;

[0030] Figure 2 This is a structural diagram of a fire warning and automatic alarm system based on a smart lamp pole provided by one embodiment of the present invention;

[0031] Figure 3 It is a structural diagram of a computer device according to an embodiment of the present invention. DETAILED DESCRIPTION

[0032] In the following description, specific details such as specific system structures and techniques are provided for purposes of illustration rather than limitation to facilitate a thorough understanding of the embodiments of the present application. However, it will be apparent to those skilled in the art that the present application may be implemented in other embodiments without these specific details. In other cases, detailed descriptions of well-known systems, devices, circuits, and methods are omitted to avoid obscuring the description of the present application with unnecessary detail.

[0033] It should be understood that when used in the present specification and the appended claims, the term "comprising" indicates the presence of described features, integers, steps, operations, elements and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components and / or collections thereof.

[0034] It will also be understood that the term "and / or" used in this specification and the appended claims refers to and includes any and all possible combinations of one or more of the associated listed items.

[0035] As used in this specification and the appended claims, the term "if" can be interpreted as "when" or "upon" or "in response to determining" or "in response to detecting," depending on the context. Similarly, the phrase "if it is determined" or "if [described condition or event] is detected" can be interpreted as meaning "upon determination" or "in response to determining" or "upon detection of [described condition or event]" or "in response to detecting [described condition or event]," depending on the context.

[0036] In addition, in the description of the present application specification and the appended claims, the terms "first", "second", "third", etc. are only used to distinguish the descriptions and cannot be understood as indicating or implying relative importance.

[0037] References to "one embodiment" or "some embodiments" in this specification mean that a particular feature, structure, or characteristic described in conjunction with that embodiment is included in one or more embodiments of the present application. Thus, phrases such as "in one embodiment," "in some embodiments," "in other embodiments," and "in other embodiments" appearing in various places in this specification do not necessarily refer to the same embodiment, but rather mean "one or more but not all embodiments," unless otherwise specifically emphasized. The terms "including," "comprising," "having," and variations thereof all mean "including but not limited to," unless otherwise specifically emphasized.

[0038] In the embodiments of the present application, the execution subject of the process includes a terminal device, which includes but is not limited to: a server, a computer, a smart phone, a tablet computer, and other devices capable of executing the method disclosed in the present application. Figure 1 A flow chart of a fire warning and automatic alarm method based on a smart lamp pole disclosed in one embodiment of the present invention is shown, and is described in detail as follows:

[0039] S101 , establishing an environmental parameter and smoke refractive index mapping model based on the difference in smoke refractive index changes in different environments, wherein the environmental parameter and smoke refractive index mapping model is used to determine a first smoke refractive index range corresponding to different environments.

[0040] In this embodiment, a high-definition camera and infrared sensor are installed on the top of the smart lamppost to capture smoke images and initially detect its presence. Temperature and humidity sensors, anemometers, and particle concentration monitors are also included to collect real-time environmental parameter data. A smoke refractive index measurement module is integrated to measure the refractive index of smoke samples using optical principles. When the infrared sensor detects possible smoke, it triggers the smoke refractive index measurement module to collect data. Simultaneously, the environmental parameter sensor records ambient temperature, humidity, wind speed, and particle concentration. The collected data undergoes preprocessing, including denoising, outlier detection, and removal, to ensure data quality.

[0041] Leveraging historical and experimental data, a mapping relationship is established between environmental parameters (temperature, humidity, wind speed, and particulate matter concentration) and smoke refractive index. Machine learning algorithms (such as decision trees, random forests, or deep learning models) are used for model training to learn how smoke refractive index varies under different environmental conditions. This model considers the impact of multiple environmental parameters on smoke refractive index, enabling the system to maintain robust early warning performance across diverse climates and geographic locations, enhancing its environmental adaptability. By integrating smoke refractive index measurement with environmental parameter monitoring, smart light poles can more accurately identify fire smoke, reduce false alarms and missed alarms, and improve fire warning accuracy.

[0042] In some embodiments, in the above step S101, establishing the environmental parameter and smoke refractive index mapping model based on the difference in smoke refractive index under different environments specifically includes:

[0043] Acquire a first smoke image dataset, perform environmental classification on the first smoke image dataset using a random forest algorithm, and determine an environmental category of each smoke image in the first smoke image dataset;

[0044] For each smoke image of the environmental category, extract corresponding environmental parameters and smoke refractive index, and establish a first mapping relationship between the environmental parameters and the smoke refractive index, wherein the environmental parameters include humidity, temperature, wind speed, and particle concentration;

[0045] According to the first mapping relationship of all smoke images in each environment category, an environment parameter and smoke refractive index mapping model is established for each environment category.

[0046] In this embodiment, a smoke image captured by a smart lamp pole camera is obtained, and the smoke image is subjected to denoising, contrast enhancement, and brightness adjustment processing to obtain an optimized smoke image dataset. Based on the optimized smoke image dataset, a random forest algorithm is used to perform environmental classification on the smoke image to determine the environmental category of the smoke image. For the smoke image of the environmental category, environmental parameters such as wind speed, atmospheric humidity, temperature, and particle concentration are extracted, and a mapping relationship between environmental parameters and refractive index is established through multivariate linear regression to obtain a calculation formula for the smoke refractive index under each environmental category. An on-site smoke image captured in real time by a smart lamp pole camera is obtained, the environmental features of the on-site smoke image are extracted, the corresponding environmental category is matched, and the refractive index calculation formula of the environmental category is substituted to calculate the smoke refractive index value under the current environment. If the smoke refractive index value exceeds the preset error range, an environmental compensation coefficient is introduced to correct the smoke refractive index value to obtain the final refractive index standard value range.

[0047] Specifically, based on the smoke images obtained by the smart lamp pole camera, the images are preprocessed, including denoising, contrast enhancement, brightness adjustment and other operations, to obtain an optimized smoke image dataset, and the shape features, edge features, density distribution, color information and texture features of the smoke are extracted from the dataset. For the optimized smoke image dataset, the random forest algorithm is used to classify smoke images in different environments, and the smoke images are divided into multiple environmental categories, such as indoor, outdoor, industrial area, residential area, etc., and each category is assigned a corresponding environmental label. For the smoke images divided into environmental categories, environmental-related parameters are extracted, including wind speed, atmospheric humidity, temperature, particle concentration, etc., and a mapping relationship between environmental parameters and refractive index is established. The calculation formula of the smoke refractive index under each environmental category is obtained through multivariate linear regression, where the refractive index R=a wind speed + b atmospheric humidity + c temperature + d particle concentration + e, and a, b, c, d, e are regression coefficients. Using the established mapping relationship between environment and refractive index, the researchers extracted environmental features from live smoke images captured by smart lamppost cameras in real time, matched them to the corresponding environmental category, and substituted them into the refractive index calculation formula for that environmental category to calculate the smoke refractive index value for the current environment. The standard range of values ​​was then determined within a preset error range δ, defined as [R-δ, R+δ]. To address the significant variations in smoke refractive index across different environments, an environmental compensation factor k was introduced. The value of k was pre-set based on the environmental category, and the calculated refractive index value was multiplied by k to obtain the final standard refractive index range. The raw smoke images captured by the smart lamppost cameras were preprocessed using Gaussian filtering for denoising, histogram equalization for contrast enhancement, and gamma correction for brightness adjustment to generate an optimized smoke image dataset. From this dataset, smoke features were extracted: shape features (such as roundness and area ratio), edge features (such as Canny edge detection results), density distribution (such as grayscale value statistical histograms), color information (such as mean and variance in the HSV color space), and texture features (such as energy and contrast in the gray-level co-occurrence matrix). The random forest algorithm uses these features to classify smoke images into environmental categories, setting the number of trees to 100 and the maximum depth to 10. A classification model is trained. The classification results categorize smoke images into environmental categories, such as indoor, outdoor, industrial, and residential areas, with each category assigned a unique numerical identifier. For each environmental category, relevant parameters such as light intensity (lux), atmospheric humidity (%), temperature (°C), and particle concentration (μg / m³) are extracted. A mapping between these environmental parameters and refractive index is established using multivariate linear regression, with the regression coefficients calculated using the least squares method. For example, for outdoor environments, the refractive index calculation formula might be R = 0.0002 light intensity + 0.001 atmospheric humidity + 0.005 temperature + 0.0001 particle concentration + 1.0. For real-time smoke images, environmental features are extracted and matched to the most similar environmental category, which is then substituted into the corresponding refractive index calculation formula.Assuming the calculated refractive index value is 1.33 and the preset error range δ is 0.02, the determined standard value range is [1.31, 1.35]. To account for differences in refractive index variation in different environments, an environmental compensation factor k is introduced, such as k = 1.0 indoors, k = 0.95 outdoors, and k = 1.05 in industrial areas. The calculated refractive index value is multiplied by k to correct it, resulting in the final refractive index standard value range. For example, the corrected range for outdoor environments is [1.24, 1.28].

[0048] S102, establish a template library of combustible material types and smoke refractive index, determine the combustible material type based on the smoke image captured by the camera of the smart lamp pole, obtain the corresponding second standard smoke refractive index range based on the determined combustible material type, and optimize the environmental parameters and smoke refractive index mapping model based on the second standard smoke refractive index range.

[0049] In this embodiment, samples of common combustible materials on the market are first collected, including but not limited to wood, paper, plastic, cloth, etc. Under controlled conditions (such as constant temperature and humidity), professional smoke generation equipment and a refractive index meter are used to measure and record the refractive index range of the smoke produced when each combustible material is burned. The collected data is organized into a library of combustible material type and smoke refractive index templates. Each template contains the combustible material name, smoke image characteristics, and the corresponding standard smoke refractive index range.

[0050] When the smart light pole camera detects smoke, it immediately captures the smoke image and transmits it to the fire warning system's image processing unit. This unit uses image recognition algorithms (such as convolutional neural networks (CNNs)) to analyze the smoke image and extract image features (such as color, texture, and shape). This extracted image feature information is then compared with the combustible material type and templates in the smoke refractive index template library. The most likely combustible material type is determined by calculating feature similarity or applying classification algorithms (such as support vector machines (SVMs) or K-nearest neighbor (KNNs). Based on the determined combustible material type, the corresponding standard smoke refractive index range is retrieved from the template library and used as the second standard smoke refractive index range.

[0051] The second standard smoke refractive index range is used as a reference for comparison and analysis with the predictions from the environmental parameter-smoke refractive index mapping model. If significant differences are found, model parameters are adjusted or new environmental variables are introduced to more accurately reflect the variations in smoke refractive index when different combustibles burn. Through continuous iterative optimization, the model's prediction accuracy and robustness are improved.

[0052] In this embodiment, by identifying the type of combustible material and obtaining the corresponding standard smoke refractive index range, the system can more accurately determine the source and nature of the smoke, reduce false alarm and missed alarm rates, and improve the accuracy of fire warnings.

[0053] In some embodiments, in step S102, establishing a template library of combustible material types and smoke refractive indices specifically includes:

[0054] Acquire smoke images generated by the combustion of different combustibles to form a second smoke image dataset;

[0055] Determining the smoke refractive index value corresponding to each combustible material by extracting the color histogram and gray-level co-occurrence matrix features of the smoke image corresponding to each combustible material in the second smoke image dataset;

[0056] forming a smoke image feature vector based on the color histogram and gray-level co-occurrence matrix features of the smoke image corresponding to each combustible material, and associating the smoke image feature vector with the type of combustible material to form a second mapping relationship;

[0057] According to the second mapping relationship of each combustible material and the corresponding smoke refractive index value, a combustible material type and smoke refractive index template library is established.

[0058] In this embodiment, smoke images generated by the combustion of different combustibles are collected through controlled combustion experiments, the color histogram and grayscale co-occurrence matrix features of the smoke are extracted, and the measured refractive index value of the smoke is obtained using a refractive index measuring instrument to form a database of correspondences between combustible types, smoke characteristics and refractive indices.

[0059] In some embodiments, in step S102, optimizing the environmental parameter and smoke refractive index mapping model according to the second standard smoke refractive index range specifically includes:

[0060] updating the weight of the environmental parameter and smoke refractive index mapping model in real time by minimizing the difference between an output result of the environmental parameter and smoke refractive index mapping model and the second standard smoke refractive index range using a gradient descent method;

[0061] The performance of the updated environmental parameter and smoke refractive index mapping model is evaluated by k-fold cross validation. When the performance evaluation index of the environmental parameter and smoke refractive index mapping model is better than the preset threshold, the updated environmental parameter and smoke refractive index mapping model is used as the target environmental parameter and smoke refractive index mapping model.

[0062] In this embodiment, based on the on-site smoke image captured by the smart lamp pole camera, the smoke image is preprocessed to determine the smoke area and extract the feature vector of the smoke area; a random forest classification algorithm is used to match the feature vector with the features in the corresponding relationship database to identify the type of combustible material currently burning; the corresponding standard smoke refractive index range is obtained according to the type of combustible material, and the model parameters are adjusted based on the output results of the multi-parameter smoke refractive index judgment model; the accuracy and generalization ability of the model are evaluated through k-fold cross-validation, and the mean absolute error and determination coefficient are calculated. If the evaluation index is better than the preset threshold, the model is deployed in actual application.

[0063] Specifically, on-site smoke images are collected in real time from smart lamp pole cameras, and the images are preprocessed, including denoising and contrast enhancement. The image segmentation algorithm is used to determine the smoke area, and the color histogram and grayscale co-occurrence matrix features of the area are extracted to construct a feature vector. The random forest classification algorithm is used to match the preprocessed smoke image feature vector with the features in the corresponding relationship database to identify the type of combustible material currently burning and obtain the standard smoke refractive index range corresponding to the combustible material. According to the identified combustible type and the corresponding standard smoke refractive index range, combined with the output results of the multi-parameter smoke refractive index judgment model, the model parameters are dynamically adjusted through the online learning algorithm to update the refractive index judgment result. Specifically, the gradient descent method is used to minimize the difference between the model prediction value and the median of the standard range, and the model weights are updated in real time. The accuracy and generalization ability of the updated model are evaluated through k-fold cross-validation, and the mean absolute error and determination coefficient are calculated. When the evaluation index is better than the preset threshold, the updated model is deployed in practical applications. To develop a template library for combustible material types and smoke refractive index, combustion experiments were conducted on 10 common combustible materials, with each sample repeated 20 times, for a total of 200 samples. Smoke refractive index values ​​were obtained using a high-precision refractive index measurement instrument (accuracy ±0.0001), and 64-dimensional color histogram and 14-dimensional gray-level co-occurrence matrix features were extracted. A smart lamppost camera captured 1920x1080 resolution images. After Gaussian filtering for denoising and adaptive histogram equalization for contrast enhancement, the Otsu threshold segmentation algorithm was applied to identify smoke regions. Feature vectors with the same dimensions as those in the template library were extracted from the segmented smoke regions. A random forest classification algorithm employed 100 decision trees with a maximum depth of 10, identifying combustible material types through a voting mechanism. The multi-parameter smoke refractive index prediction model initially employed a five-layer fully connected neural network architecture. The input layer contained 100 nodes, including environmental parameters and image features. The hidden layers had 64, 32, 16, and 8 nodes, respectively. The output layer contained a single node representing the predicted refractive index value. During online learning, stochastic gradient descent with a learning rate of 0.01 was used, with a mini-batch update strategy of 32 samples per iteration. Model performance was evaluated using 5-fold cross-validation, calculating the mean absolute error (MAE) and coefficient of determination (R²). When MAE < 0.001 and R² > 0.95, the updated model was considered to perform well and was subsequently deployed in production. The entire processing flow was completed on the edge computing unit, with an average processing time of 100 milliseconds per frame. This enabled real-time and accurate assessment of the refractive index of smoke generated by different combustible materials and dynamic model optimization.

[0064] S103 , identifying the smoke image captured by the camera of the smart lamp pole according to the environmental parameters and the smoke refractive index mapping model, and determining a change value of the smoke refractive index in the current environment.

[0065] In this embodiment, environmental parameter sensors (such as temperature and humidity sensors, anemometers, and particulate matter concentration monitors) integrated into smart light poles continuously monitor surrounding environmental parameters and transmit real-time data to the fire warning system. When the smart light pole's camera detects smoke, it immediately captures a high-definition smoke image and transmits it to the fire warning system's image processing unit. The image processing unit first preprocesses the captured smoke image, including steps such as noise removal, contrast enhancement, and edge detection, to better extract smoke features. An image recognition algorithm (such as a convolutional neural network (CNN)) is then used to extract features from the preprocessed smoke image, identifying key features such as smoke texture, shape, and density. Real-time environmental parameters (such as temperature, humidity, wind speed, and particulate matter concentration) are input into a trained model mapping environmental parameters to smoke refractive index. Based on the input environmental parameters, the model predicts a baseline range for smoke refractive index in the current environment. Combining features extracted from the smoke image, a specific algorithm (such as a machine learning-based regression model) is used to link the smoke features to the smoke refractive index. Based on feature changes in the smoke image, the change in the current smoke refractive index relative to the model's predicted baseline range is estimated.

[0066] S104 , establishing a smoke refractive index calibration model according to the smoke refractive index under different illumination conditions, and performing correction processing on the smoke refractive index change value under the current environment according to the smoke refractive index calibration model to obtain a corrected smoke refractive index change value.

[0067] In this example, a standard smoke source and refractive index measurement equipment were used to measure the refractive index of smoke under various lighting conditions (e.g., sunny, cloudy, dusk, and night). A high-definition camera was used to record smoke images under these lighting conditions and collect environmental parameter data (e.g., temperature, humidity, and wind speed). The collected data was preprocessed, including denoising, outlier removal, and data normalization, to ensure data quality.

[0068] Analyze how smoke refractive index changes under different lighting conditions, as well as the impact of factors such as light intensity, smoke concentration, and environmental parameters on the refractive index. Use regression analysis, machine learning algorithms (such as support vector machines (SVMs) and random forests), or deep learning models to establish a mapping between lighting conditions and smoke refractive index, creating a smoke refractive index calibration model. The model should accept current lighting conditions (such as light intensity and light source type) and a preliminary measured smoke refractive index as input and output the corrected smoke refractive index.

[0069] When the smart light pole's camera captures a smoke image, the system first uses image processing techniques to determine the current lighting conditions (e.g., through image brightness and color analysis). Simultaneously, it uses environmental parameter sensors to obtain current environmental parameter data. The lighting conditions, environmental parameters, and the initially measured smoke refractive index are input into a smoke refractive index calibration model for real-time correction. The model outputs a corrected change in the smoke refractive index, which serves as the basis for fire risk assessment. This corrected change in the smoke refractive index is compared with a preset threshold to assess the smoke's dangerousness. If the change exceeds the threshold, a fire warning signal is triggered, and an alert is sent to the fire department and relevant authorities.

[0070] In this embodiment, by establishing a smoke refractive index calibration model based on different lighting conditions, the system can correct the influence of lighting conditions on smoke refractive index measurement and improve the accuracy of the measurement results.

[0071] In some embodiments, in step S104, establishing a smoke refractive index calibration model based on smoke refractive indices under different lighting conditions specifically includes:

[0072] Obtain real-time illumination data and corresponding smoke refractive index data collected by the light intensity sensor of the smart lamp pole;

[0073] The median absolute deviation method is used to detect outliers on the real-time illumination data and to remove abnormal data points;

[0074] Classifying the real-time illumination data according to a preset fixed threshold value to obtain illumination intensity levels and corresponding time periods;

[0075] With the light intensity level as the independent variable and the smoke refractive index calibration coefficient as the dependent variable, a polynomial regression algorithm is used to establish a light intensity and smoke refractive index calibration model based on the real-time light data and the corresponding smoke refractive index data.

[0076] Furthermore, the correction process of the smoke refractive index change value in the current environment according to the smoke refractive index calibration model specifically includes:

[0077] Determine the current light intensity and current time information in the current environment based on the smoke image captured by the camera of the smart lamp pole, and determine whether the current environment is daytime or nighttime based on the current time information;

[0078] When the current environment is daytime, determining a first correction coefficient using the smoke refractive index calibration model according to the current light intensity in the current environment;

[0079] When the current environment is at night, determining a second correction coefficient using the smoke refractive index calibration model according to the current light intensity in the current environment;

[0080] The change value of the smoke refractive index in the current environment is corrected according to the first correction coefficient or the second correction coefficient.

[0081] In this embodiment, real-time light data collected by the light intensity sensor of the smart lamp pole is obtained, and the real-time light data includes a light intensity value and a corresponding timestamp; the median absolute deviation method is used to detect outliers in the light intensity data, and abnormal data points are eliminated; the light intensity data is graded according to a preset fixed threshold to obtain light intensity levels and their corresponding time periods; a polynomial regression algorithm is used to establish a smoke refractive index calibration model, and the calibration model uses light intensity as an independent variable and a smoke refractive index calibration coefficient as a dependent variable; when obtaining a smoke image from the smart lamp pole camera, the current light intensity value and time information are read synchronously; if the time information is in night mode, a night compensation factor is applied to the calibration coefficient; the calibration coefficient is applied to the output result of the environmental parameter and smoke refractive index mapping model to obtain a corrected smoke refractive index change value.

[0082] For example, a light intensity sensor is integrated into a smart light pole, collecting real-time light data every minute. The collected light intensity values ​​are associated with timestamps and stored to form a light intensity time series database. Outlier detection is performed on the light intensity data, using the median absolute deviation method to identify and remove anomalous data points. Light intensity is categorized into five levels based on preset fixed thresholds: very weak (0-50 lux), weak (51-200 lux), moderate (201-1000 lux), strong (1001-5000 lux), and very strong (>5000 lux). The time periods corresponding to each level are recorded. A polynomial regression algorithm is used to establish a smoke refractive index calibration model, with light intensity as the independent variable and the smoke refractive index calibration coefficient as the dependent variable. Light intensity and corresponding smoke refractive index data are collected for one month, sampling every hour, resulting in 720 data points for fitting the calibration function. When capturing smoke images from the smart light pole camera, the current light intensity value and time information are simultaneously read. The light intensity is substituted into the calibration function to calculate the calibration coefficient. If the current time is in night mode (e.g., 9:00 PM - 5:00 AM), an additional night compensation factor is applied. The calibration factor is multiplied by the ambient parameters and the output of the smoke refractive index mapping model to obtain the corrected smoke refractive index change. The light intensity sensor integrated into the smart light pole collects data once a minute, with a measurement range of 0-100,000 lux and an accuracy of ±1 lux. The collected data is detected for outliers using the median absolute deviation method, with a threshold of 2.5 standard deviations. Data points outside this range are marked as outliers and removed. Light intensity is categorized using preset thresholds: very weak (0-50 lux), weak (51-200 lux), moderate (201-1000 lux), strong (1001-5000 lux), and very strong (>5000 lux). The polynomial regression algorithm uses a third-order polynomial of the form y=ax³+bx²+cx+d, where x is the light intensity and y is the calibration factor. The 720 sets of data points collected within a month were fitted using the least squares method to obtain specific coefficients: a=2.3e-11, b=-1.5e-7, c=3.2e-4, d=0.98. When the light sensor reading is 2000 lux, the calibration coefficient is 1.12 when substituted into the formula. If the current time is 23:00, which is night mode, an additional night compensation factor of 1.2 is applied. The multi-parameter smoke refractive index judgment model outputs an original refractive index value of 1.35. After correction by the calibration coefficient and night compensation factor, the final refractive index judgment value is 1.35*1.12*1.2=1.8144. The entire processing process is automatically completed on the edge computing unit of the smart lamp pole, with an average response time of less than 100 milliseconds, realizing real-time and accurate judgment of the smoke refractive index under different lighting conditions.

[0083] S105 , determining a fire probability value according to the modified smoke refractive index change value, and judging whether there is a fire hazard according to the fire probability value.

[0084] In this embodiment, a fire probability assessment model is constructed based on historical fire data and the statistical relationship between changes in smoke refractive index and fire occurrence. The model should be able to receive the corrected smoke refractive index change value as input and output the corresponding fire probability value. The fire probability value can be expressed as a percentage, reflecting the likelihood of fire occurrence under current conditions. When the camera of the smart lamp pole captures the smoke image and obtains the corrected smoke refractive index change value through the above process, this value is input into the fire probability assessment model. The fire probability assessment model calculates the current fire probability value based on the input corrected smoke refractive index change value, combined with the model's internal algorithms and parameters. Other environmental parameters (such as temperature, humidity, wind speed, etc.) and other smoke characteristics (such as color, density, diffusion rate, etc.) may also be considered during the calculation process to improve the accuracy of the assessment.

[0085] The calculated fire probability value is compared with a preset threshold. If the fire probability value exceeds the threshold, a fire hazard is determined to exist and a fire warning signal is triggered. Warning signals can include audible alarms, text message notifications, app push notifications, and other methods to promptly notify relevant personnel and take countermeasures. Once a fire warning is triggered, the system immediately sends an alert to the fire department, providing detailed information such as the fire location and probability value, enabling a rapid response. The system can also collaborate with other smart city facilities (such as intelligent sprinkler systems and ventilation systems) to implement preliminary self-rescue measures if conditions permit. After the fire incident, the system should record and analyze the event to optimize the fire warning model and improve the accuracy of future warnings.

[0086] In this embodiment, by combining the corrected smoke refractive index change value with the fire probability assessment model, the system can more accurately assess the fire risk and reduce false alarms and missed alarms.

[0087] In some embodiments, in step S105, determining a fire probability value based on the modified smoke refractive index change value, and judging whether there is a fire hazard based on the fire probability value specifically include:

[0088] Calculating fire probability values ​​of preset continuous frames of smoke images using a pre-trained logistic regression model based on the corrected smoke refractive index value;

[0089] If the smoke images of the preset consecutive frames all exceed the preset fire probability threshold, it is judged that there is a fire hazard and the alarm mechanism is triggered.

[0090] In this embodiment, a pre-trained logistic regression model is used to calculate the fire probability value based on the smoke refractive index. If the fire probability value exceeds a preset threshold, a fire hazard is determined. Time series analysis is performed on multiple consecutive frames of image data. If the fire probability values ​​for each consecutive frame exceed the preset threshold, an alarm mechanism is triggered. An image segmentation algorithm is used to determine the center coordinates of the smoke region, which are then transmitted to the firefighter's terminal as the possible location of the fire source.

[0091] Based on the fire warning signals and possible fire source locations, the smart light poles will upload suspected fire images and environmental data to the city's fire protection big data platform for secondary confirmation. At the same time, the platform will link with nearby fire hydrants and drone IoT devices to prompt nearby people to disperse and carry out further fire reconnaissance and firefighting operations, forming an integrated fire warning and automatic alarm system for city-level smart light pole fire protection.

[0092] Specifically, after detecting a fire warning signal, the smart light pole packages the suspected fire image, environmental data, and fire source coordinates into a standard JSON-formatted data packet, which is uploaded in real time to the city fire protection big data platform via a high-speed wireless network. This data packet simultaneously triggers nearby smart light poles to enter an alert state. Upon receiving the data packet, the city fire protection big data platform immediately activates a convolutional neural network to conduct a secondary verification of the suspected fire image. It then combines historical fire data with information about surrounding buildings for comprehensive analysis to determine the fire's authenticity and severity. If the secondary verification confirms a real fire, the platform dynamically calculates the impact area based on the fire's severity and sends an opening command to smart fire hydrants within that area. It also dispatches the nearest firefighting drone to the fire source for real-time monitoring and data collection. The platform uses the A* algorithm combined with real-time traffic information to generate an evacuation route map. Voice prompts and LED displays on nearby smart light poles guide nearby residents to evacuate in an orderly manner. Fire reports are also sent to the fire command center to provide decision-making support for subsequent firefighting operations. The city-wide smart light pole fire protection integrated system utilizes a shared database for information exchange and distributed computing to improve response speed, forming a collaborative working mechanism that automates the entire process from fire detection to firefighting operations. Upon detecting a fire warning signal, the smart light pole immediately packages a 1920x1080 resolution image of the suspected fire, along with environmental data such as temperature, humidity, and wind speed, and the centimeter-accurate coordinates of the fire source, into a JSON-formatted data packet. This data packet is then uploaded to the city's firefighting big data platform via a 4G / 5G network at 10Mbps. Upon receiving the data packet, the platform uses a ResNet50 convolutional neural network to perform secondary validation of the image, achieving 98% accuracy. The platform also utilizes a random forest algorithm to assess fire severity, using the past five years of historical fire data and information on buildings within a 500-meter radius. A score of 60 or higher is considered a real fire, and the platform dynamically calculates the impact radius based on the score, where r = 10sqrt(score) meters. An open command is then sent to all smart fire hydrants within this radius, maintaining the water pressure at 0.4MPa at each hydrant. Simultaneously, the nearest DJI Mavic 3 firefighting drone is dispatched to the fire source at a speed of 15m / s, hovering at an altitude of 100m for monitoring. The platform uses algorithm A combined with real-time traffic information to generate evacuation route maps, calculating up to eight alternative routes. Voice prompts are played at 90 decibels via the nearest smart light poles, and evacuation instructions scroll across LED screens. The city-wide smart light pole fire protection integrated system utilizes a distributed Kafka message queue for data sharing, achieving processing latency of less than 100ms and ensuring a full response time of less than 10 seconds.

[0093] Reference Figure 2 An embodiment of the present invention provides a fire warning and automatic alarm system 2 based on a smart lamp pole, and the system 2 specifically includes:

[0094] A first processing module 201 is configured to establish a mapping model between environmental parameters and smoke refractive index based on the variation of smoke refractive index in different environments, wherein the mapping model is used to determine a first smoke refractive index range corresponding to different environments;

[0095] A second processing module is configured to establish a template library of combustible material types and smoke refractive indexes, determine the combustible material type based on the smoke image captured by the camera of the smart lamp pole, obtain a corresponding second standard smoke refractive index range based on the determined combustible material type, and optimize the environmental parameter and smoke refractive index mapping model based on the second standard smoke refractive index range;

[0096] a third processing module, configured to identify the smoke image captured by the camera of the smart lamp pole according to the environmental parameters and the smoke refractive index mapping model, and determine a change value of the smoke refractive index in the current environment;

[0097] a fourth processing module, configured to establish a smoke refractive index calibration model based on the smoke refractive index under different lighting conditions, and perform correction processing on the smoke refractive index change value under the current environment based on the smoke refractive index calibration model to obtain a corrected smoke refractive index change value;

[0098] The fifth processing module is used to determine a fire probability value according to the modified smoke refractive index change value, and determine whether there is a fire hazard according to the fire probability value.

[0099] It is understandable that if Figure 1 The contents of the embodiment of the fire warning and automatic alarm method based on the smart light pole shown in the figure are applicable to the embodiment of the fire warning and automatic alarm system based on the smart light pole. The functions specifically implemented by the embodiment of the fire warning and automatic alarm system based on the smart light pole are the same as those in the embodiment of the fire warning and automatic alarm system based on the smart light pole. Figure 1 The fire warning and automatic alarm method based on the smart lamp pole shown in the embodiment is the same as that of the embodiment shown in the embodiment, and the beneficial effects achieved are the same as those of the embodiment shown in the embodiment. Figure 1 The beneficial effects achieved by the fire warning and automatic alarm method embodiment based on the smart lamp pole shown are also the same.

[0100] It should be noted that the information interaction, execution process and other contents between the above-mentioned systems are based on the same concept as the embodiment of the method of the present invention. Their specific functions and technical effects can be found in the method embodiment part and will not be repeated here.

[0101] Those skilled in the art can clearly understand that, for the convenience and brevity of description, only the division of the above-mentioned functional units and modules is used as an example for illustration. In actual applications, the above-mentioned functions can be distributed and completed by different functional units and modules as needed, that is, the internal structure of the system can be divided into different functional units or modules to complete all or part of the functions described above. The functional units and modules in the embodiment can be integrated into one processing unit, or each unit can exist physically alone, or two or more units can be integrated into one unit. The above-mentioned integrated unit can be implemented in the form of hardware or in the form of software functional units. In addition, the specific names of the functional units and modules are only for the convenience of distinguishing each other, and are not used to limit the scope of protection of this application. The specific working process of the units and modules in the above-mentioned system can refer to the corresponding process in the aforementioned method embodiment, and will not be repeated here.

[0102] Reference Figure 3 The embodiment of the present invention further provides a computer device 3, comprising: a memory 302 and a processor 301 and a computer program 303 stored in the memory 302. When the computer program 303 is executed on the processor 301, the fire warning and automatic alarm method based on the smart lamp pole as described in any one of the above methods is implemented.

[0103] The computer device 3 may be a desktop computer, a notebook computer, a PDA, a cloud server or other computing devices. The computer device 3 may include, but is not limited to, a processor 301 and a memory 302. Those skilled in the art will understand that Figure 3 This is merely an example of the computer device 3 and does not constitute a limitation on the computer device 3 . The computer device 3 may include more or fewer components than shown in the figure, or a combination of certain components, or different components. For example, the computer device 3 may also include input and output devices, network access devices, etc.

[0104] The processor 301 may be a central processing unit (CPU), or other general-purpose processors, digital signal processors (DSP), application-specific integrated circuits (ASIC), field-programmable gate arrays (FPGA), or other programmable logic devices, discrete gate or transistor logic devices, or discrete hardware components. A general-purpose processor may be a microprocessor or any conventional processor.

[0105] In some embodiments, the memory 302 may be an internal storage unit of the computer device 3, such as a hard drive or memory of the computer device 3. In other embodiments, the memory 302 may also be an external storage device of the computer device 3, such as a plug-in hard drive, a Smart Media Card (SMC), a Secure Digital (SD) card, a flash memory card, etc. equipped on the computer device 3. Furthermore, the memory 302 may include both an internal storage unit of the computer device 3 and an external storage device. The memory 302 is used to store an operating system, application programs, a boot loader, data, and other programs, such as the program code of the computer program. The memory 302 may also be used to temporarily store data that has been output or is about to be output.

[0106] An embodiment of the present invention further provides a computer-readable storage medium having a computer program stored thereon. When the computer program is executed by a processor, the fire warning and automatic alarm method based on a smart lamp pole as described in any one of the above methods is implemented.

[0107] In this embodiment, if the integrated unit is implemented as a software functional unit and sold or used as a standalone product, it can be stored in a computer-readable storage medium. Based on this understanding, the present application can implement all or part of the process steps in the above-mentioned method embodiments by using a computer program to instruct the relevant hardware. The computer program can be stored in a computer-readable storage medium. When executed by a processor, the computer program can implement the steps of each of the above-mentioned method embodiments. The computer program includes computer program code, which can be in source code form, object code form, executable file, or some intermediate form. The computer-readable medium can include at least: any entity or device capable of carrying computer program code to a camera / terminal device, recording medium, computer memory, read-only memory (ROM), random access memory (RAM), electric carrier signals, telecommunication signals, and software distribution media. Examples include USB flash drives, removable hard drives, magnetic disks, or optical disks. In some jurisdictions, based on legislation and patent practice, computer-readable media cannot be electric carrier signals or telecommunication signals.

[0108] In the above embodiments, the description of each embodiment has its own focus. For parts that are not described or recorded in detail in a certain embodiment, reference can be made to the relevant description of other embodiments.

[0109] Those skilled in the art will appreciate that the units and algorithm steps of each example described in conjunction with the embodiments disclosed herein can be implemented in electronic hardware, or a combination of computer software and electronic hardware. Whether these functions are performed in hardware or software depends on the specific application and design constraints of the technical solution. Professional and technical personnel can use different methods to implement the described functions for each specific application, but such implementation should not be considered beyond the scope of this application.

[0110] In the embodiments disclosed in the present application, it should be understood that the disclosed devices / terminal equipment and methods can be implemented in other ways. For example, the device / terminal equipment embodiments described above are merely schematic. For example, the division of the modules or units is merely a logical function division. In actual implementation, there may be other division methods, such as multiple units or components can be combined or integrated into another system, or some features can be ignored or not executed. Another point is that the mutual coupling or direct coupling or communication connection shown or discussed can be through some interfaces, indirect coupling or communication connection of devices or units, which can be electrical, mechanical or other forms.

[0111] The units described as separate components may or may not be physically separate, and the components shown as units may or may not be physical units, that is, they may be located in one place or distributed across multiple network units. Some or all of these units may be selected to achieve the purpose of this embodiment according to actual needs.

Claims

1. A fire warning and automatic alarm method based on smart lamp poles, characterized in that: The method specifically includes: Based on the differences in the refractive index of smoke in different environments, a mapping model between environmental parameters and smoke refractive index is established, wherein the mapping model between environmental parameters and smoke refractive index is used to determine the first smoke refractive index range corresponding to different environments; The establishment of a mapping model between environmental parameters and smoke refractive index based on the difference in smoke refractive index under different environments specifically includes: Acquire a first smoke image dataset, perform environmental classification on the first smoke image dataset using a random forest algorithm, and determine an environmental category of each smoke image in the first smoke image dataset; For each smoke image of the environmental category, extract corresponding environmental parameters and smoke refractive index, and establish a first mapping relationship between the environmental parameters and the smoke refractive index, wherein the environmental parameters include humidity, temperature, wind speed, and particle concentration; Establishing an environmental parameter and smoke refractive index mapping model for each environmental category based on the first mapping relationship of all smoke images under each environmental category; Establish a template library of combustible material types and smoke refractive indexes, determine the combustible material type based on the smoke image captured by the smart lamp pole camera, obtain the corresponding second standard smoke refractive index range based on the determined combustible material type, and optimize the environmental parameter and smoke refractive index mapping model based on the second standard smoke refractive index range; The establishment of a template library of combustible material types and smoke refractive index specifically includes: Acquire smoke images generated by the combustion of different combustibles to form a second smoke image dataset; Determining the smoke refractive index value corresponding to each combustible material by extracting the color histogram and gray-level co-occurrence matrix features of the smoke image corresponding to each combustible material in the second smoke image dataset; forming a smoke image feature vector based on the color histogram and gray-level co-occurrence matrix features of the smoke image corresponding to each combustible material, and associating the smoke image feature vector with the type of combustible material to form a second mapping relationship; According to the second mapping relationship of each combustible material and the corresponding smoke refractive index value, a combustible material type and smoke refractive index template library is established; Identify the smoke image captured by the camera of the smart lamp pole according to the environmental parameters and the smoke refractive index mapping model, and determine the change value of the smoke refractive index in the current environment; Establishing a smoke refractive index calibration model based on the smoke refractive index under different lighting conditions, and correcting the smoke refractive index change value under the current environment according to the smoke refractive index calibration model to obtain a corrected smoke refractive index change value; A fire probability value is determined based on the modified smoke refractive index change value, and whether there is a fire hazard is determined based on the fire probability value.

2. The method according to claim 1, characterized in that Optimizing the environmental parameter and smoke refractive index mapping model according to the second standard smoke refractive index range specifically includes: updating the weight of the environmental parameter and smoke refractive index mapping model in real time by minimizing the difference between an output result of the environmental parameter and smoke refractive index mapping model and the second standard smoke refractive index range using a gradient descent method; The performance of the updated environmental parameter and smoke refractive index mapping model is evaluated by k-fold cross validation. When the performance evaluation index of the environmental parameter and smoke refractive index mapping model is better than the preset threshold, the updated environmental parameter and smoke refractive index mapping model is used as the target environmental parameter and smoke refractive index mapping model.

3. The method according to claim 1, characterized in that The smoke refractive index calibration model is established according to the smoke refractive index under different lighting conditions, specifically including: Obtain real-time illumination data and corresponding smoke refractive index data collected by the light intensity sensor of the smart lamp pole; The median absolute deviation method is used to detect outliers on the real-time illumination data and to remove abnormal data points; Classifying the real-time illumination data according to a preset fixed threshold value to obtain illumination intensity levels and corresponding time periods; With the light intensity level as the independent variable and the smoke refractive index calibration coefficient as the dependent variable, a polynomial regression algorithm is used to establish a light intensity and smoke refractive index calibration model based on the real-time light data and the corresponding smoke refractive index data.

4. The method according to claim 3, characterized in that The correction process of the smoke refractive index change value under the current environment according to the smoke refractive index calibration model specifically includes: Determine the current light intensity and current time information in the current environment based on the smoke image captured by the camera of the smart lamp pole, and determine whether the current environment is daytime or nighttime based on the current time information; When the current environment is daytime, determining a first correction coefficient using the smoke refractive index calibration model according to the current light intensity in the current environment; When the current environment is at night, determining a second correction coefficient using the smoke refractive index calibration model according to the current light intensity in the current environment; The change value of the smoke refractive index in the current environment is corrected according to the first correction coefficient or the second correction coefficient.

5. The method according to claim 1, wherein Determining a fire probability value based on the modified smoke refractive index change value, and judging whether there is a fire hazard based on the fire probability value specifically includes: Calculating fire probability values ​​of preset continuous frames of smoke images using a pre-trained logistic regression model based on the corrected smoke refractive index value; If the smoke images of the preset consecutive frames all exceed the preset fire probability threshold, it is judged that there is a fire hazard and the alarm mechanism is triggered.

Images