Multi-scene-oriented AI fire safety monitoring and automatic early warning system

Abstract

The invention discloses a multi-scene-oriented AI fire safety monitoring and automatic early warning system, and belongs to the technical field of fire monitoring, and the system comprises a data collection module which is used for periodically collecting environment factor data influencing a fire hazard, and obtaining a basic data set; the environment evaluation module is used for calculating an environment evaluation value by utilizing the environment factor data in the basic data set; the monitoring mode selection module is used for selecting an intelligent monitoring mode, a multi-source fusion monitoring mode and a sensor monitoring mode according to the environmental evaluation value; the fire-fighting evaluation module is used for evaluating the possibility of fire occurrence according to the selected mode; and the early warning module is used for generating and sending early warning information according to the evaluation result. According to the invention, by adopting an intelligent data processing and analysis technology, a monitoring mode can be flexibly selected and the possibility of fire occurrence can be accurately evaluated according to different environmental factors and scene characteristics, so that early warning information can be sent out in time, and a powerful guarantee is provided for life and property safety of people.

Description

Technical Field

[0001] The present invention relates to the field of fire monitoring technology, and in particular to an AI fire safety monitoring and automatic early warning system for multiple scenarios. Background Art

[0002] With the continuous development of science and technology, the demand for intelligence, automation and precision in the field of fire safety monitoring is increasing. Traditional fire safety monitoring systems can no longer meet the needs of modern society, especially in complex environments such as large public places, industrial parks, and residential areas. An efficient, accurate and intelligent fire safety monitoring system is needed to protect people's lives and property.

[0003] At present, traditional fire safety monitoring systems often rely on a single monitoring method, such as smoke detectors or temperature sensors. These systems may show good monitoring effects in specific scenarios, but in complex and changing environments, their monitoring effects and warning accuracy are often limited; single monitoring methods are easily affected by environmental factors, resulting in frequent false alarms or missed alarms; secondly, traditional systems lack intelligent processing capabilities and cannot conduct in-depth analysis of the collected data, making it difficult to accurately assess the possibility of fire; finally, the warning method of traditional systems is relatively single, and can often only provide simple alarm information, which cannot meet the warning needs in multiple scenarios. Summary of the invention

[0004] To solve the above problems, the present invention provides an AI fire safety monitoring and automatic early warning system for multiple scenarios. By adopting intelligent data processing and analysis technology, it can flexibly select monitoring modes and accurately assess the possibility of fire according to different environmental factors and scene characteristics, thereby issuing early warning information in a timely manner, providing strong protection for people’s lives and property safety.

[0005] The above objectives can be achieved through the following solutions:

[0006] An AI fire safety monitoring and automatic early warning system for multiple scenarios includes a data acquisition module for periodically collecting data on environmental factors that affect fire to obtain a basic data set; an environmental assessment module for calculating an environmental assessment value using the environmental factor data in the basic data set; a monitoring mode selection module for selecting an intelligent monitoring mode, a multi-source fusion monitoring mode, and a sensor monitoring mode according to the size of the environmental assessment value; a fire assessment module for assessing the possibility of fire according to the selected mode; and an early warning module for generating and sending early warning information according to the assessment result.

[0007] Optionally, the periodic collection of data on environmental factors that affect fire to obtain a basic data set includes: collecting environmental temperature data, humidity data, wind speed data and sunshine data to obtain a first environmental data set; performing missing value processing and anomaly detection processing on the first environmental data set to obtain a second environmental data set; performing denoising processing on the second environmental data set to obtain a third environmental data set; and normalizing the third environmental data set to obtain a basic data set.

[0008] Optionally, the calculating of the environmental assessment value by using the environmental factor data in the basic data set includes: using the basic data set to establish an environmental evaluation function for characterizing the environmental assessment value, and for the environmental assessment value K,

[0009] K=ω1*W+ω2*F+ω3*R-ω4*S,

[0010] In the formula, ω1 is the weight coefficient of temperature, W is the numerical data corresponding to the temperature in the basic data set, ω2 is the weight coefficient of wind speed, F is the numerical data corresponding to the wind speed in the basic data set, ω3 is the weight coefficient of sunshine, r is the numerical data corresponding to sunshine in the basic data set, ω4 is the weight coefficient of humidity, and S is the numerical data corresponding to humidity in the basic data set; the basic data set is substituted into the environmental evaluation function to calculate the current environmental assessment value.

[0011] Optionally, the selection of the intelligent monitoring mode, the multi-source fusion monitoring mode and the sensor monitoring mode according to the size of the environmental assessment value includes: judging whether the environmental assessment value is greater than a preset first threshold; if the environmental assessment value is greater than the first threshold, switching to the intelligent monitoring mode; if the environmental assessment value is less than or equal to the first threshold, judging whether the environmental assessment value is greater than a preset second threshold; if the environmental assessment value is greater than the second threshold, switching to the multi-source fusion monitoring mode; if the environmental assessment value is less than or equal to the second threshold, switching to the sensor monitoring mode.

[0012] Optionally, the fire assessment module includes: an intelligent monitoring and assessment unit, a multi-source fusion assessment unit and a sensor assessment unit; wherein the intelligent monitoring and assessment unit is used to assess the possibility of fire using a neural network model; the multi-source fusion assessment unit is used to analyze multi-source data and assess the possibility of fire based on the analysis results; the sensor assessment unit is used to analyze the possibility of fire through infrared sensors and smoke sensors.

[0013] Optionally, the use of a neural network model to assess the likelihood of a fire includes: acquiring historical environmental data corresponding to when a fire occurred and when a fire did not occur, to obtain a historical data set; preprocessing the data in the historical data set, extracting fire characteristic data from the preprocessed historical data set, to obtain a historical characteristic data set; marking the fire characteristic data of a fire that occurred in the historical characteristic data set, to obtain a training data set; using the fire characteristic data as input and whether a fire occurred as output, establishing and training a neural network model using the training data set, to obtain a fire occurrence prediction model.

[0014] Optionally, the use of a neural network model to assess the likelihood of a fire also includes: collecting and preprocessing environmental data; extracting fire feature data from the preprocessed environmental data to obtain an input data set; inputting the input data set into the fire occurrence prediction model to obtain an occurrence result; when the occurrence result is that there is a fire, determining that the likelihood of a fire is high; and when the occurrence result is that there is no fire, determining that the likelihood of a fire is low.

[0015] Optionally, the analysis of multi-source data and assessment of the possibility of fire based on the analysis results include: collecting smoke data, temperature data, light data, heat source data and image data of the environment; preprocessing and normalizing the smoke data and temperature data of the environment to obtain a numerical parameter set; analyzing the light data and heat source data of the environment, and combining them with the image data to determine whether there is an abnormal light source; and digitizing the determination result to obtain a state parameter value.

[0016] Optionally, analyzing the multi-source data and evaluating the possibility of fire occurrence according to the analysis results includes: establishing an evaluation function for characterizing a fire evaluation value; substituting the numerical parameter set and the state parameter value into the evaluation function to calculate a fire evaluation value, and for the fire evaluation value P, there is

[0017]

[0018] Wherein, θ is the weight coefficient corresponding to the state parameter value, Y is the state parameter value, is the weight coefficient corresponding to the smoke data, X1 is the data corresponding to the smoke data in the numerical parameter set, is the weight coefficient corresponding to the temperature data, X2 is the data corresponding to the temperature data in the numerical parameter set; it is determined whether the fire assessment value is greater than a preset third threshold; if so, it is determined that the possibility of fire is high; if not, it is determined that the possibility of fire is low.

[0019] Optionally, the analysis of the possibility of fire through infrared sensors and smoke sensors includes: using infrared sensors to collect heat source data, and generating a first warning signal when the heat source data reaches a preset fourth threshold; using smoke sensors to collect smoke concentration data, and generating a second warning signal when the smoke concentration data reaches a preset fifth threshold; determining whether the first warning signal and the second warning signal exist at the same time; if so, determining that the possibility of fire is high; if not, determining that the possibility of fire is low.

[0020] Compared with the prior art, the present invention has the following advantages:

[0021] 1. The present invention realizes the automation and intelligence of fire safety monitoring by adopting intelligent data processing and analysis technology; the system can automatically collect environmental data, evaluate fire risks, select monitoring modes and generate early warning information without manual intervention, which greatly improves the monitoring efficiency and accuracy;

[0022] 2. The system design takes into account the needs of various scenarios. By flexibly selecting monitoring modes (intelligent monitoring mode, multi-source fusion monitoring mode and sensor monitoring mode), the system can adapt to the fire monitoring needs in different scenarios and provide more accurate early warning services;

[0023] 3. Using the environmental assessment module and the fire assessment module, the system can comprehensively consider a variety of environmental factors (such as temperature, humidity, wind speed, sunshine, etc.) and data sources (such as sensor data, historical environmental data, etc.), and accurately assess the possibility of fire through advanced algorithms and models (such as neural network models, multi-source data fusion algorithms, etc.). This helps to reduce false alarms and missed alarms and improve the accuracy and reliability of early warnings.

[0024] Other features and advantages of the present invention will be described in the following description, and partly become apparent from the description, or understood by practicing the present invention. The purpose and other advantages of the present invention can be realized and obtained by the structures pointed out in the description, claims and drawings. BRIEF DESCRIPTION OF THE DRAWINGS

[0025] In order to more clearly illustrate the embodiments of the present invention or the technical solutions in the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings described below are some embodiments of the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0026] Figure 1 It is a framework diagram of an AI fire safety monitoring and automatic early warning system for multiple scenarios in an embodiment of the present invention.

[0027] Figure 2 It is a flowchart of the operation of an AI fire safety monitoring and automatic early warning system for multiple scenarios in an embodiment of the present invention.

[0028] Figure 3 It is an execution flow chart of an AI fire safety monitoring and automatic early warning system for multiple scenarios in an embodiment of the present invention.

[0029] Figure 4 It is a structural diagram of an AI fire safety monitoring and automatic early warning system for multiple scenarios according to an embodiment of the present invention. DETAILED DESCRIPTION

[0030] In order to make the purpose, technical solution and advantages of the embodiments of the present invention clearer, the technical solution in the embodiments of the present invention will be clearly and completely described below in conjunction with the drawings in the embodiments of the present invention. Obviously, the described embodiments are part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0031] Reference Figure 1 An embodiment of the present invention proposes an AI fire safety monitoring and automatic early warning system for multiple scenarios. By adopting intelligent data processing and analysis technology, it can flexibly select monitoring modes and accurately assess the possibility of fire according to different environmental factors and scene characteristics, thereby issuing early warning information in a timely manner, providing strong protection for people's lives and property safety.

[0032] The system of this embodiment specifically includes:

[0033] The data collection module is used to periodically collect data on environmental factors that affect fire and obtain a basic data set;

[0034] An environmental assessment module, used to calculate an environmental assessment value using the environmental factor data in the basic data set;

[0035] A monitoring mode selection module is used to select intelligent monitoring mode, multi-source fusion monitoring mode and sensor monitoring mode according to the size of the environmental assessment value;

[0036] A fire assessment module, used to assess the possibility of fire according to the selected mode;

[0037] The early warning module is used to generate and send early warning information based on the evaluation results.

[0038] Specifically, the early warning module generates and sends early warning information based on the evaluation results of the fire assessment module so that relevant personnel can take timely measures; including the possibility of fire, the location of the fire, the severity of the fire, etc.; early warning information can be sent through SMS, email, APP push and other methods.

[0039] For example, Figure 2 As shown, data on environmental factors that affect fire are periodically collected to obtain a basic data set, and the environmental factor data in the basic data set are used to calculate the environmental assessment value; the intelligent monitoring mode, multi-source fusion monitoring mode and sensor monitoring mode are selected according to the size of the environmental assessment value; the possibility of fire is evaluated according to the selected mode, and early warning information is generated and sent according to the evaluation results; the automation and intelligence of fire safety monitoring are realized by adopting intelligent data processing and analysis technology; the system can automatically collect environmental data, evaluate fire risks, select monitoring modes and generate early warning information without human intervention, which greatly improves the monitoring efficiency and accuracy.

[0040] Optionally, the periodic collection of environmental factor data affecting fire to obtain a basic data set includes:

[0041] Collecting environmental temperature data, humidity data, wind speed data and sunshine data to obtain a first environmental data set;

[0042] Specifically, the data acquisition module collects environmental data in real time through various sensors (such as temperature sensors, humidity sensors, wind speed sensors and sunshine sensors) arranged in the monitoring area; these data reflect the environmental conditions in the monitoring area and are the basis for subsequent analysis.

[0043] Performing missing value processing and anomaly detection processing on the first environmental data set to obtain a second environmental data set;

[0044] Specifically, due to sensor failure or data transmission problems, the first environmental data set may contain missing values ​​or outliers; in order to improve data quality, these data need to be processed; for missing data, interpolation methods (such as linear interpolation, spline interpolation, etc.) or prediction methods based on historical data can be used to fill in the missing data; outliers can be detected and removed by setting reasonable thresholds or using machine learning algorithms (such as isolation forests, DBSCAN, etc.).

[0045] Performing denoising processing on the second environmental data set to obtain a third environmental data set;

[0046] Specifically, the data collected by the sensor may be interfered by various noises, such as electromagnetic interference, mechanical vibration, etc.; in order to obtain more accurate environmental data, the second environmental data set needs to be denoised; denoising methods include filtering methods (such as low-pass filtering, high-pass filtering, band-pass filtering, etc.), wavelet transform method, Kalman filtering method, etc.

[0047] The third environment data set is normalized to obtain a basic data set.

[0048] Specifically, since the data collected by different sensors may have different dimensions and value ranges, in order to facilitate subsequent data analysis and processing, the third environment data set needs to be normalized; normalization methods include min-max normalization, Z-score normalization, etc.; through normalization, the data can be converted into numerical values ​​under a unified scale, which is convenient for subsequent data analysis and processing.

[0049] Optionally, the calculating the environmental assessment value by using the environmental factor data in the basic data set includes:

[0050] Using the basic data set, an environmental evaluation function is established to characterize the environmental evaluation value. For the environmental evaluation value K,

[0051] K=ω1*W+ω2*F+ω3*R-ω4*S,

[0052] Wherein, ω1 is the weight coefficient of temperature, W is the numerical data corresponding to the temperature in the basic data set, ω2 is the weight coefficient of wind speed, F is the numerical data corresponding to the wind speed in the basic data set, ω3 is the weight coefficient of sunshine, R is the numerical data corresponding to sunshine in the basic data set, ω4 is the weight coefficient of humidity, and S is the numerical data corresponding to humidity in the basic data set;

[0053] Substitute the basic data set into the environmental evaluation function to calculate the current environmental evaluation value.

[0054] Specifically, the environmental assessment module is responsible for calculating the environmental assessment value using the environmental factor data in the basic data set; this process aims to provide a basis for subsequent monitoring mode selection and fire assessment by quantifying the impact of environmental factors on fire risk.

[0055] Exemplarily, it is assumed that the basic data set has been normalized and the numerical range is between 0 and 1; the numerical data corresponding to the temperature in the basic data set is 0.7, the numerical data corresponding to the wind speed in the basic data set is 0.5, the numerical data corresponding to the sunshine in the basic data set is 0.8, and the numerical data corresponding to the humidity in the basic data set is 0.3; it is assumed that the weight coefficient of temperature is 0.4, the weight coefficient of wind speed is 0.2, the weight coefficient of sunshine is 0.3, and the weight coefficient of humidity is 0.1; these values ​​are substituted into the environmental evaluation function, and K=0.4×0.7+0.2×0.5+0.3×0.8-0.1×0.3=0.59 is calculated; therefore, the current environmental assessment value is 0.59, which can be used to judge the possibility or risk level of fire under current environmental conditions according to the preset threshold, thereby guiding subsequent monitoring mode selection and fire assessment work.

[0056] Alternatively, if Figure 3 As shown, the selection of intelligent monitoring mode, multi-source fusion monitoring mode and sensor monitoring mode according to the size of the environmental assessment value includes:

[0057] Determining whether the environmental assessment value is greater than a preset first threshold;

[0058] If the environmental assessment value is greater than the first threshold, switching to intelligent monitoring mode;

[0059] Specifically, if the environmental assessment value is greater than the first threshold, it indicates that the fire risk is high under the current environmental conditions and more intelligent and comprehensive monitoring methods are needed; at this time, the monitoring mode selection module will switch to the intelligent monitoring mode, which may use advanced algorithms and models (such as neural network models) to conduct a comprehensive analysis of multiple environmental factors to more accurately assess the possibility of fire.

[0060] If the environmental assessment value is less than or equal to the first threshold, determining whether the environmental assessment value is greater than a preset second threshold;

[0061] If the environmental assessment value is greater than the second threshold, switching to a multi-source fusion monitoring mode;

[0062] Specifically, if the environmental assessment value is less than or equal to the first threshold but greater than the preset second threshold, it indicates that the fire risk under the current environmental conditions is moderate, and a more comprehensive but not overly complex monitoring method is needed; at this time, the monitoring mode selection module will switch to the multi-source fusion monitoring mode, which integrates data from different sources (such as smoke data, temperature data, light data, heat source data, and image data, etc.) and evaluates the possibility of fire through the fusion analysis of multi-source data.

[0063] If the environmental assessment value is less than or equal to the second threshold, the sensor monitoring mode is switched.

[0064] Specifically, if the environmental assessment value is less than or equal to the second threshold, it indicates that the fire risk under the current environmental conditions is low, and a simpler and more direct monitoring method can be adopted; at this time, the monitoring mode selection module will switch to the sensor monitoring mode, which mainly relies on physical sensors such as infrared sensors and smoke sensors to monitor parameters such as heat sources and smoke concentrations in the environment. When these parameters exceed the preset threshold, an early warning signal is generated.

[0065] Specifically, the monitoring mode selection module is responsible for flexibly selecting the appropriate monitoring mode according to the size of the environmental assessment value; this process aims to ensure that the most effective monitoring means can be used under different environmental conditions through intelligent decision-making, thereby improving the accuracy and timeliness of fire warnings.

[0066] Exemplarily, assuming that the first threshold is 0.6 and the second threshold is 0.3, through the previous environmental assessment calculation, the current environmental assessment value is 0.5; since the environmental assessment value 0.5 is less than the first threshold 0.6, it is then determined whether the environmental assessment value is greater than the second threshold. Since the environmental assessment value 0.5 is greater than the first threshold 0.3, the condition for switching the multi-source fusion monitoring mode is met; according to the above judgment result, the monitoring mode selection module will switch to the multi-source fusion monitoring mode and start integrating data from different sources for fire risk assessment; this process ensures that the most appropriate monitoring mode can be used under different environmental conditions, thereby improving the accuracy and timeliness of fire warnings.

[0067] Optionally, as shown in the figure, the fire assessment module includes: an intelligent monitoring assessment unit, a multi-source fusion assessment unit and a sensor assessment unit; wherein,

[0068] The intelligent monitoring and evaluation unit is used to evaluate the possibility of fire occurrence using a neural network model;

[0069] Specifically, the intelligent monitoring and evaluation unit uses advanced neural network models, such as deep learning networks, to learn and train large amounts of historical fire data and environmental factor data. Through model learning, it can capture the complex relationship between environmental factors and fire occurrence, so that when new data is input, it can accurately assess the possibility of fire occurrence.

[0070] The multi-source fusion evaluation unit is used to analyze the multi-source data and evaluate the possibility of fire occurrence according to the analysis results;

[0071] Specifically, the multi-source fusion assessment unit integrates information from different sensors and data sources, such as smoke concentration, temperature changes, light changes, heat source images, etc.; through the fusion analysis of these multi-source data, it can more comprehensively assess the possibility of fire and improve the accuracy of the assessment.

[0072] The sensor evaluation unit is used to analyze the possibility of fire occurrence through infrared sensors and smoke sensors.

[0073] Specifically, the sensor evaluation unit mainly relies on physical sensors, such as infrared sensors and smoke sensors, to monitor parameters such as heat sources and smoke concentrations in the environment; when these parameters exceed preset thresholds, it is considered that the possibility of a fire is high.

[0074] Specifically, the fire assessment module can flexibly respond to fire risk assessment needs under various environmental conditions through different combinations and applications of intelligent monitoring and assessment units, multi-source fusion assessment units and sensor assessment units; the intelligent monitoring and assessment unit is suitable for high-precision assessment in complex environments; the multi-source fusion assessment unit is suitable for medium-precision assessment that requires comprehensive consideration of multiple factors; the sensor evaluation unit is suitable for rapid warning in simple environments; this modular design makes the fire assessment system more flexible, efficient and reliable.

[0075] Optionally, the method of evaluating the possibility of fire occurrence by using a neural network model includes:

[0076] Obtain historical environmental data corresponding to fire occurrence and fire non-occurrence to obtain a historical data set;

[0077] Specifically, historical environmental data corresponding to fires and fires not occurring are collected to construct the historical data set required for training the neural network model; historical environmental data should include various environmental factors that may affect the occurrence of fires, such as temperature, humidity, wind speed, precipitation, vegetation coverage, terrain conditions, etc.

[0078] Preprocessing the data in the historical data set, extracting fire characteristic data from the preprocessed historical data set to obtain a historical characteristic data set;

[0079] Specifically, improve data quality to make data more suitable for training neural network models; remove duplicate, missing or outlier data through data cleaning; convert data of different dimensions to the same dimension through data normalization so that the model can learn better; select features that have a significant impact on the occurrence of fire from the original data to form a feature data set.

[0080] Specifically, fire feature data is extracted from the preprocessed historical data set to train the neural network model; statistical methods, data mining techniques or domain knowledge are used to identify features related to fire occurrence.

[0081] The fire feature data of the fires that occurred in the historical feature data set are marked to obtain a training data set;

[0082] Specifically, the fire feature data of fires in the historical feature dataset are labeled so that the model can distinguish between fires and non-fires; binary labels are usually used, such as 1 for fires and 0 for non-fires.

[0083] Taking fire characteristic data as input and whether a fire occurs as output, a neural network model is established and trained using the training data set to obtain a fire occurrence prediction model.

[0084] Specifically, according to the complexity of the problem and the scale of the data, a suitable neural network structure is selected, such as a multi-layer perceptron (MLP), a convolutional neural network (CNN), etc.; the neural network model is trained using a training data set, and the prediction error is minimized by adjusting the parameters of the model (such as weights and biases); the trained model is evaluated using a validation data set to verify its prediction performance; the model is optimized based on the evaluation results, such as adjusting the network structure, hyperparameters, or using regularization methods to prevent overfitting; after training and evaluation, a neural network model that can accurately predict the occurrence of fire is obtained; the model is applied to the actual environment, the possibility of fire is predicted based on real-time input environmental data, and corresponding preventive or response measures are taken.

[0085] Optionally, the method of evaluating the possibility of fire occurrence by using a neural network model further includes:

[0086] Collect environmental data and pre-process it;

[0087] Specifically, environmental data should include various factors related to the occurrence of fire, such as temperature, humidity, wind speed, smoke concentration, infrared thermal imaging, etc.; collect environmental data in real time through various sensors deployed in the monitoring area (such as temperature sensors, humidity sensors, wind speed sensors, smoke sensors, infrared cameras, etc.); remove duplicate, missing or outlier data from the collected data; normalize or standardize the data to make data of different dimensions comparable.

[0088] Extract fire characteristic data from the preprocessed environmental data to obtain an input data set;

[0089] Specifically, it is necessary to clarify which characteristics are closely related to fire, such as temperature, smoke concentration, light changes, etc., and ensure that these characteristics can accurately reflect the occurrence of fire.

[0090] Inputting the input data set into the fire occurrence prediction model to obtain an occurrence result;

[0091] When the occurrence result is that there is a fire, the possibility of the fire occurring is judged to be high;

[0092] When the occurrence result is no fire, the possibility of fire occurrence is judged to be low.

[0093] Specifically, the input data set is input into the fire occurrence prediction model. The model will process the input data according to the learned knowledge and rules and output the prediction results. When the prediction result is that there is a fire, the possibility of the fire is judged to be high and corresponding emergency measures need to be taken immediately. When the prediction result is that there is no fire, the possibility of the fire is judged to be low and routine monitoring can continue.

[0094] For example, suppose that in a forest fire monitoring system, monitoring is required through an intelligent monitoring mode; first, environmental data is collected in real time through temperature sensors, humidity sensors, wind speed sensors, smoke sensors and infrared cameras deployed in the forest; the collected raw data is cleaned, outliers are removed, and the data is normalized to scale the temperature, humidity, wind speed and other data to between 0 and 1; feature data such as abnormal temperature increase (such as exceeding a certain threshold), increased smoke concentration, and abnormal hot spots in infrared thermal images are extracted from the preprocessed data; the extracted feature data is organized into the form of an input data set, such as {temperature: 0.8, humidity: 0.4, wind speed: 0.6, smoke concentration: 0.9, infrared thermal image abnormality: 1}; the input data set is input into the trained neural network model, and the model processes the input data and outputs the prediction result; assuming that the prediction result of the model is that there is a fire, the possibility of the fire is determined to be high, and the fire warning system is immediately activated to notify relevant personnel to carry out emergency handling; if the prediction result of the model is that there is no fire, the possibility of the fire is determined to be low, and routine monitoring continues.

[0095] Optionally, analyzing the multi-source data and evaluating the possibility of fire occurrence according to the analysis results includes:

[0096] Collect environmental smoke data, temperature data, light data, heat source data and image data;

[0097] Specifically, the smoke concentration in the environment is collected through a smoke sensor, the temperature of the environment is collected through a temperature sensor, the light intensity or change of the environment is collected through a light sensor or a camera, the heat source distribution in the environment is collected through an infrared sensor or a thermal imaging camera, and the real-time image of the environment is collected through a camera.

[0098] Preprocess and normalize the smoke data and temperature data of the environment to obtain a numerical parameter set;

[0099] Specifically, duplicate, missing or outlier data are removed, and the smoke concentration and temperature data are scaled to a uniform range (e.g., between 0 and 1) for numerical comparison and calculation, thereby obtaining a numerical parameter set including normalized smoke concentration and temperature data.

[0100] Analyze the ambient light data and heat source data, and combine them with the image data to determine whether there is an abnormal light source;

[0101] Specifically, by comprehensively analyzing light, heat source and image data, abnormal light sources that may exist in the environment, such as light emitted by flames or high-temperature objects, can be identified; changes in light intensity can be observed to identify whether there is a sudden increase in or abnormal fluctuation in light; heat sources in the environment can be identified through infrared sensors or thermal imaging cameras, and their distribution and changes can be observed; by processing real-time images, it can be confirmed whether there are abnormal light sources that match the light and heat source data, such as flames, smoke or high-temperature objects; it can be determined whether there are abnormal light sources in the environment, and their location and range can be determined.

[0102] The judgment result is digitized to obtain the state parameter value.

[0103] Specifically, the judgment result is converted into a numerical form for quantitative analysis and comparison; if there is an abnormal light source, a corresponding numerical value is assigned (such as 1 for the presence of an abnormal light source, and 0 for the absence of an abnormal light source), and a state parameter value is obtained to indicate the presence of an abnormal light source in the environment.

[0104] Optionally, analyzing the multi-source data and evaluating the possibility of fire occurrence according to the analysis results includes:

[0105] Establishing an evaluation function for characterizing fire evaluation values;

[0106] Substitute the numerical parameter set and the state parameter value into the evaluation function to calculate the fire evaluation value. For the fire evaluation value P,

[0107]

[0108] Wherein, θ is the weight coefficient corresponding to the state parameter value, Y is the state parameter value, is the weight coefficient corresponding to the smoke data, X1 is the data corresponding to the smoke data in the numerical parameter set, is the weight coefficient corresponding to the temperature data, and X2 is the data corresponding to the temperature data in the numerical parameter set;

[0109] Specifically, a mathematical model (i.e., an evaluation function) is established to integrate multi-source data to quantitatively evaluate the possibility of fire; smoke data, temperature data and other related data are collected and preprocessed to obtain a set of numerical parameters; based on the analysis results of the abnormal light source, the state parameter value is obtained, which is 0 or 1; these values ​​are substituted into the evaluation function to calculate the fire assessment value.

[0110] Determining whether the fire assessment value is greater than a preset third threshold;

[0111] If yes, the probability of fire is judged to be high;

[0112] If not, the possibility of fire is judged to be low.

[0113] Specifically, the possibility of fire is determined by comparing the fire assessment value with a preset threshold; a third threshold is set as a standard for determining the possibility of fire; if the fire assessment value is greater than the third threshold, the possibility of fire is determined to be high; otherwise, the possibility of fire is determined to be low.

[0114] For example, assume that in a warehouse fire monitoring system, the above evaluation function is used to evaluate the possibility of fire; the smoke concentration collected by the smoke sensor is 25ppm, which is normalized to 0.5 (assuming the maximum concentration is 50ppm); the temperature collected by the temperature sensor is 40 degrees Celsius, which is normalized to 0.8 (assuming the maximum temperature is 50 degrees Celsius); through light and heat source analysis, combined with image data, it is judged that there is an abnormal light source, so the state parameter value is 1; assuming that the weight coefficient corresponding to the state parameter value is 0.6, the weight coefficient corresponding to the smoke data is 0.3, and the weight coefficient corresponding to the temperature data is 0.1; then the fire assessment value P = 0.6×1+0.3×0.5+0.1×0.8=0.83, assuming that the third threshold is 0.7, the fire assessment value 0.83 is greater than the third threshold 0.7, so the possibility of fire is judged to be high; based on this judgment result, the system can trigger the early warning mechanism to notify the warehouse management personnel to check and handle to prevent the occurrence of fire.

[0115] Optionally, analyzing the possibility of fire occurrence by using an infrared sensor and a smoke sensor includes:

[0116] Collecting heat source data using an infrared sensor, and generating a first warning signal when the heat source data reaches a preset fourth threshold;

[0117] Specifically, the infrared sensor can detect heat sources in the environment, such as flames or high-temperature objects, which is one of the early signs of a fire; the infrared sensor continuously monitors the intensity of infrared radiation in the environment and converts it into electrical signals for processing; a fourth threshold is preset. When the heat source data collected by the infrared sensor reaches or exceeds this threshold, it is considered that there may be a fire risk. Once the heat source data reaches the fourth threshold, the infrared sensor generates a first warning signal.

[0118] Using a smoke sensor to collect smoke concentration data, when the smoke concentration data reaches a preset fifth threshold, a second warning signal is generated;

[0119] Specifically, the smoke sensor can detect the smoke concentration in the environment, and smoke is another important sign of a fire. The smoke sensor inhales air samples, analyzes the concentration of smoke particles therein, and converts them into electrical signals. A fifth threshold is preset. When the smoke concentration data collected by the smoke sensor reaches or exceeds this threshold, it is considered that there may be a fire. Once the smoke concentration data reaches the fifth threshold, the smoke sensor generates a second warning signal.

[0120] Determining whether the first warning signal and the second warning signal exist at the same time;

[0121] If yes, the probability of fire is judged to be high;

[0122] If not, the possibility of fire is judged to be low.

[0123] Specifically, the accuracy of fire judgment is improved by integrating the warning signals of infrared sensors and smoke sensors; the system continuously monitors the existence of the first warning signal and the second warning signal; if the two warning signals exist at the same time, that is, both the infrared sensor and the smoke sensor detect signs of fire, then the possibility of fire is determined to be high; otherwise, the possibility of fire is determined to be low.

[0124] It should be noted that the electrical connection between the above-mentioned units does not necessarily mean direct connection of the lines, and the indirect connection mode can be applied to the embodiments of the present invention as long as the purpose of the present invention is achieved. The above is only an exemplary embodiment of the present invention and cannot be used to limit the scope of the present invention.

[0125] That is, any equivalent changes and modifications made according to the teachings of the present invention are still within the scope of the present invention. After considering the disclosure of the specification and the truth of practice, those skilled in the art will easily think of other embodiments of the present invention. This application is intended to cover any variation, use or adaptive change of the present invention, which follows the general principles of the present invention and includes common knowledge or customary technical means in the art that are not described in the present invention.

Claims

1. An AI fire safety monitoring and automatic early warning system for multiple scenarios, characterized in that: The system comprises: The data collection module is used to periodically collect data on environmental factors that affect fire and obtain a basic data set; An environmental assessment module, used to calculate an environmental assessment value using the environmental factor data in the basic data set; A monitoring mode selection module is used to select intelligent monitoring mode, multi-source fusion monitoring mode and sensor monitoring mode according to the size of the environmental assessment value; A fire assessment module, used to assess the possibility of fire according to the selected mode; The early warning module is used to generate and send early warning information based on the evaluation results.

2. According to claim 1, the multi-scenario AI fire safety monitoring and automatic early warning system is characterized in that: The periodic collection of environmental factor data affecting fire to obtain a basic data set includes: Collecting environmental temperature data, humidity data, wind speed data and sunshine data to obtain a first environmental data set; Performing missing value processing and anomaly detection processing on the first environmental data set to obtain a second environmental data set; Performing denoising processing on the second environmental data set to obtain a third environmental data set; The third environment data set is normalized to obtain a basic data set.

3. According to the multi-scenario AI fire safety monitoring and automatic early warning system of claim 1, it is characterized in that: The calculating of the environmental assessment value by using the environmental factor data in the basic data set comprises: Using the basic data set, an environmental evaluation function is established to characterize the environmental evaluation value. For the environmental evaluation value K, K=ω1*W+ω2*F+ω3*R-ω4*S, Wherein, ω1 is the weight coefficient of temperature, W is the numerical data corresponding to temperature in the basic data set, ω2 is the weight coefficient of wind speed, F is the numerical data corresponding to wind speed in the basic data set, ω3 is the weight coefficient of sunshine, r is the numerical data corresponding to sunshine in the basic data set, ω4 is the weight coefficient of humidity, and s is the numerical data corresponding to humidity in the basic data set; Substitute the basic data set into the environmental evaluation function to calculate the current environmental evaluation value.

4. According to claim 1, the multi-scenario AI fire safety monitoring and automatic early warning system is characterized in that: The selection of intelligent monitoring mode, multi-source fusion monitoring mode and sensor monitoring mode according to the size of the environmental assessment value includes: Determining whether the environmental assessment value is greater than a preset first threshold; If the environmental assessment value is greater than the first threshold, switching to intelligent monitoring mode; If the environmental assessment value is less than or equal to the first threshold, determining whether the environmental assessment value is greater than a preset second threshold; If the environmental assessment value is greater than the second threshold, switching to a multi-source fusion monitoring mode; If the environmental assessment value is less than or equal to the second threshold, the sensor monitoring mode is switched.

5. According to claim 1, the multi-scenario AI fire safety monitoring and automatic early warning system is characterized in that: The fire assessment module includes: an intelligent monitoring assessment unit, a multi-source fusion assessment unit and a sensor assessment unit; wherein, The intelligent monitoring and evaluation unit is used to evaluate the possibility of fire occurrence using a neural network model; The multi-source fusion evaluation unit is used to analyze the multi-source data and evaluate the possibility of fire occurrence according to the analysis results; The sensor evaluation unit is used to analyze the possibility of fire occurrence through infrared sensors and smoke sensors.

6. The multi-scenario AI fire safety monitoring and automatic early warning system according to claim 5 is characterized in that: The use of a neural network model to assess the possibility of fire occurrence includes: Obtain historical environmental data corresponding to fire occurrence and fire non-occurrence to obtain a historical data set; Preprocessing the data in the historical data set, extracting fire characteristic data from the preprocessed historical data set to obtain a historical characteristic data set; The fire feature data of the fires that occurred in the historical feature data set are marked to obtain a training data set; Taking fire characteristic data as input and whether a fire occurs as output, a neural network model is established and trained using the training data set to obtain a fire occurrence prediction model.

7. The multi-scenario AI fire safety monitoring and automatic early warning system according to claim 6 is characterized in that: The use of a neural network model to assess the possibility of fire also includes: Collect environmental data and pre-process it; Extract fire characteristic data from the preprocessed environmental data to obtain an input data set; Inputting the input data set into the fire occurrence prediction model to obtain an occurrence result; When the occurrence result is that there is a fire, the possibility of the fire occurring is judged to be high; When the occurrence result is no fire, the possibility of fire occurrence is determined to be low.

8. The multi-scenario AI fire safety monitoring and automatic early warning system according to claim 5 is characterized in that: The analysis of multi-source data and assessment of the possibility of fire occurrence based on the analysis results include: Collect environmental smoke data, temperature data, light data, heat source data and image data; Preprocess and normalize the smoke data and temperature data of the environment to obtain a numerical parameter set; Analyze the ambient light data and heat source data, and combine them with the image data to determine whether there is an abnormal light source; The judgment result is digitized to obtain the state parameter value.

9. The multi-scenario AI fire safety monitoring and automatic early warning system according to claim 8 is characterized in that: The analysis of multi-source data and assessment of the possibility of fire occurrence based on the analysis results include: Establishing an evaluation function for characterizing fire evaluation values; Substitute the numerical parameter set and the state parameter value into the evaluation function to calculate the fire evaluation value. For the fire evaluation value P, Wherein, θ is the weight coefficient corresponding to the state parameter value, Y is the state parameter value, is the weight coefficient corresponding to the smoke data, X1 is the data corresponding to the smoke data in the numerical parameter set, is the weight coefficient corresponding to the temperature data, and X2 is the data corresponding to the temperature data in the numerical parameter set; Determining whether the fire assessment value is greater than a preset third threshold; If yes, the probability of fire is judged to be high; If not, the possibility of fire is judged to be low.

10. The multi-scenario AI fire safety monitoring and automatic early warning system according to claim 5 is characterized in that: The possibility of fire occurrence analyzed by infrared sensors and smoke sensors includes: Collecting heat source data using an infrared sensor, and generating a first warning signal when the heat source data reaches a preset fourth threshold; Using a smoke sensor to collect smoke concentration data, when the smoke concentration data reaches a preset fifth threshold, a second warning signal is generated; Determining whether the first warning signal and the second warning signal exist at the same time; If yes, the probability of fire is judged to be high; If not, the possibility of fire is judged to be low.

Images