Smart city fire monitoring and early warning method and system based on Internet of Things
By calculating the smoke surge index and continuous index, and combining the prediction network to identify the fire source, the problems of false alarms and underreports of existing systems are solved, and accurate identification and timely response to fires are achieved.
Patent Information
- Application Number
- CN202411471400.0
- Authority / Receiving Office
- CN · China
- Patent Type
- Applications(China)
- Current Assignee / Owner
- Filing Date
- 2024-10-22
- Publication Date
- 2024-11-19
- Estimated Expiration
- Not applicable · inactive patent
AI Technical Summary
The existing smart city fire monitoring and early warning system based on the Internet of Things is prone to accidentally triggering alarms due to non-fire situations such as cooking smoke, bathroom steam or cigarette smoke, resulting in the neglect of real fire alarms and cannot be detected in time when the smoke is isolated or in high airflow areas. .
By obtaining smoke concentration data in real time, calculating smoke surge index and continuous index, dynamically adjusting monitoring time, combining preset prediction network to predict smoke dispersion points, identifying fire sources and issuing early warnings.
Significantly reduce false alarms, improve the accuracy and reliability of fire alarms, ensure timely identification of fires and optimize the allocation of fire resources.
Smart Images

Figure CN118982893A_ABST
Abstract
Description
Technical Field
[0001] The present invention relates to the field of fire monitoring and early warning, and more specifically, to a method and system for fire monitoring and early warning in a smart city based on the Internet of Things. Background Art
[0002] The development of IoT technology has provided strong support for the construction of smart cities, enabling them to collect large amounts of data through smart devices and sensors and optimize urban management. Among them, fire safety is particularly important in smart cities. The fire monitoring and early warning system has been upgraded to use real-time data analysis to quickly identify fire risks, reduce false alarms and improve sensitivity. It is convenient for the system to issue early warnings in a timely manner, providing key support for early detection and response to fires, thereby effectively protecting people's lives and property, and reflecting the important role of smart cities in improving urban safety and quality of life.
[0003] The existing Chinese patent application document with publication number CN114971409A discloses a smart city fire monitoring and early warning method and system based on the Internet of Things. The method is executed by a management platform, including: obtaining monitoring data collected by an object platform through a sensor network platform, the monitoring data including smoke data, temperature data, drone image data, and manual inspection data, and the manual inspection data is obtained based on the manual inspection time interval; determining the fire risk level based on the monitoring data; in response to the fire risk level meeting the preset conditions, issuing an alarm to the user platform through the service platform; in response to the fire risk level not meeting the preset conditions, determining the manual inspection time interval based on the fire risk level, and the manual inspection time interval is sent to the service platform and / or the user platform.
[0004] In the above invention, the fire risk level is obtained through platform data, and the interval of manual inspection is adjusted to improve the rational allocation of resources. However, false alarms are common at present. The alarm may be triggered by cooking smoke, bathroom steam or cigarette smoke instead of fire. This may not only cause dissatisfaction among residents, but also lead to the neglect of real fire alarms. If the sensitivity is limited, when the smoke is isolated in a closed space or far away from the alarm, it means that some types of smoke or fire scenes may not be detected in time. Environmental factors will also have an impact, such as high airflow areas interfering with the spread of smoke and affecting the detection ability of the alarm. Summary of the invention
[0005] To solve the problem that the alarm may be falsely triggered by cooking smoke, bathroom steam or cigarette smoke instead of fire situation, resulting in the neglect of real fire alarm, the present invention provides solutions in the following aspects.
[0006] In the first aspect, a smart city fire monitoring and early warning method based on the Internet of Things includes: acquiring concentration data of smoke concentration at each monitoring point in real time; identifying concentration anomalies based on concentration data and a preset safety threshold, and dynamically acquiring monitoring time based on the concentration anomalies, wherein the concentration anomalies are inversely proportional to the monitoring time, extracting a sequence of concentration anomalies within the monitoring time, and calculating a smoke surge index to determine a suspected fire source; calculating a continuous index of the smoke surge index of the monitoring point, multiplying the continuous index product of all moments as the degree of continuity of the monitoring point, judging the ignition point according to the degree of continuity, and taking the minimum value point of the non-zero smoke surge index as the monitoring starting point; predicting the monitoring starting point based on a preset prediction network, obtaining the probability of a smoke diffusion point, predicting the distribution of smoke at a future time point, and issuing an early warning alarm based on the prediction result; wherein the continuous index satisfies the following relationship: ; In the formula, Indicates Moment The continuous index of each monitoring point, Indicates Moment The smoke surge index at each monitoring point is Indicates Moment The smoke surge index at each monitoring point.
[0007] The effects are as follows: by accurately identifying abnormal concentration values and dynamic monitoring time, the system can distinguish non-fire situations such as cooking smoke, bathroom steam or cigarette smoke, reducing false alarms; using the smoke surge index and continuity index, the system can identify real fire situations and improve the accuracy of alarms; the system can calculate the smoke surge index and continuity index in a timely manner to quickly identify the source of fire; by reducing false alarms, the reliability of alarms in real fires is improved, avoiding missing the opportunity to extinguish early fires due to neglect.
[0008] Preferably, the abnormal concentration values include: In response to the detected concentration data being greater than or equal to a preset safety threshold, the concentration abnormality value of the smoke is the difference between the concentration data and the preset safety threshold. Conversely, if the concentration data is less than the preset safety threshold, the concentration abnormality value is 0.
[0009] Preferably, the monitoring time satisfies the following relationship: ; In the formula, Indicates the monitoring time. Indicates The concentration abnormal value at the moment, Indicates the standard monitoring time, Represents a natural constant.
[0010] Preferably, the smoke surge index includes: Get the monitoring time window length , get the Time to The concentration anomaly value sequence at a certain moment is taken as the target moment. The ratio between the concentration anomaly value at the target moment and the moment of monitoring time before the target moment is calculated. The standard deviation of the concentration anomaly value sequence is calculated to measure the degree of dispersion of each data point in the sequence relative to the mean value. The standard deviation is exponentially decayed, and the product of the ratio and the exponentially decayed standard deviation is taken as the smoke surge index at the target moment.
[0011] The effect is as follows: by comprehensively considering the sudden changes and fluctuations in smoke concentration, the method can more accurately identify real fire situations and reduce false alarms; by dynamically adjusting the monitoring time window, the method can adapt to different stages of fire development and can effectively monitor from the initial stage to the fierce stage; by distinguishing between natural fluctuations in smoke concentration and abnormal changes caused by fire, the method reduces false alarms caused by non-fire situations such as cooking smoke, bathroom steam or cigarette smoke.
[0012] Preferably, the degree of continuity includes: In response to the continuity degree being 1, the monitoring points are continuous and in a smoke diffusion channel; conversely, when the continuity degree is 0, the monitoring points are not on the smoke diffusion channel, and multiple diffusion channels can be obtained, each diffusion channel includes multiple monitoring points, labels are set for the monitoring points on the diffusion channels, and the monitoring point matrix is obtained by arranging the monitoring point positions.
[0013] The effect is: by obtaining the continuity between the monitoring points through the degree of continuity, it can be determined that the smoke is diffusing through a continuous path, which helps to accurately identify the propagation path of the smoke, which helps to identify multiple independent smoke diffusion channels, each of which may involve multiple monitoring points. By identifying the smoke diffusion channels, the system can more accurately predict the diffusion path and range of the smoke, and provide more reliable information for fire emergency response and personnel evacuation.
[0014] Preferably, predicting the monitoring starting point based on a preset prediction network includes: Smoke points are artificially created at preset positions, and a preset number of real-time monitoring matrices of smoke points in a diffusion process are collected as a training set for a training, and prediction matrices at multiple moments are obtained. The prediction matrices are put into a preset prediction network, and in response to a loss function being less than a preset value or reaching a preset number of training times, a smoke diffusion point prediction network is obtained.
[0015] The effect is: by artificially creating smoke points at preset locations and collecting data, a more accurate smoke dispersion model can be obtained, thereby improving the prediction accuracy of smoke dispersion points. By predicting smoke dispersion points, firefighting and rescue resources can be more effectively allocated and unnecessary waste of resources can be reduced.
[0016] Preferably, the smoke dispersion point prediction network is a convolutional neural network, including: an input layer: receiving raw input data; a convolution layer: applying a learnable filter to slide on the input data to capture local features; a pooling layer: used to reduce the feature space dimension and reduce the amount of calculation; a fully connected layer: extracting features and mapping them to the final output; an output layer: outputting category probabilities through an activation function; The real-time monitoring matrix is put into the smoke diffusion point prediction network to obtain the prediction matrix for the next moment in the future. The monitoring point corresponding to 1 point in the prediction matrix is the smoke diffusion point. When the smoke diffusion point is greater than the preset value, a fire alarm is issued.
[0017] The effect is: by obtaining a real-time monitoring point matrix, the location of the smoke dispersion point at a certain moment in the future can be obtained, so that a rapid warning can be given in the early stages of a fire. At the same time, the smoke dispersion channel can provide key information for personnel at the fire scene and assist in planning escape routes. By understanding the areas where smoke may spread, people can be guided to avoid high-risk areas and choose safer escape routes.
[0018] In the second aspect, a smart city fire monitoring and early warning system based on the Internet of Things includes: a processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the above-mentioned smart city fire monitoring and early warning method based on the Internet of Things is implemented.
[0019] The present invention has the following effects: 1. Through the dynamic monitoring time and calculation of the smoke surge index, the system can distinguish between fire smoke and other non-fire smoke sources (such as cooking smoke, bathroom steam, etc.), thereby significantly reducing false alarms. At the same time, the continuous degree of evaluation helps to identify real fire events, reduce the possibility of missed alarms, and ensure timely response when a real fire occurs.
[0020] 2. The present invention predicts smoke diffusion points by using convolutional neural networks, and the system can predict the distribution of smoke at future time points. This prediction capability helps optimize the allocation of firefighting resources, guide personnel evacuation, and formulate more effective emergency response strategies. In addition, by predicting multiple smoke diffusion channels, the system can provide a more comprehensive perspective for fire rescue and improve the overall emergency response efficiency and effectiveness. BRIEF DESCRIPTION OF THE DRAWINGS
[0021] By reading the following detailed description with reference to the accompanying drawings, the above and other objects, features and advantages of the exemplary embodiments of the present invention will become readily understood. In the accompanying drawings, several embodiments of the present invention are shown in an exemplary and non-limiting manner, and the same or corresponding reference numerals represent the same or corresponding parts, wherein: Figure 1 It is a method flow chart of steps S1 to S4 in the smart city fire monitoring and early warning method based on the Internet of Things in an embodiment of the present invention.
[0022] Figure 2 It is a bird's-eye view schematic diagram of the layout of monitoring points on each floor in the smart city fire monitoring and early warning system based on the Internet of Things in an embodiment of the present invention.
[0023] Figure 3 It is a structural block diagram of a smart city fire monitoring and early warning system based on the Internet of Things according to an embodiment of the present invention. DETAILED DESCRIPTION
[0024] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions 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 those skilled in the art without creative work are within the scope of protection of the present invention.
[0025] The specific embodiments of the present invention are described in detail below with reference to the accompanying drawings.
[0026] Reference Figure 1 The fire monitoring and early warning method for smart cities based on the Internet of Things includes steps S1 to S4, which are as follows: S1: Obtain smoke concentration data of each monitoring point in real time.
[0027] For example, refer to Figure 2 , set up monitoring point A in the building to collect real-time smoke concentration data.
[0028] Further explanation: when a fire occurs, combustibles are gradually ignited, the fire source grows from small to large, and the smoke concentration is also increasing. The smoke continues to diffuse outward with ventilation. The closer to the fire source, the higher the smoke concentration. There is a first monitoring point around, and a surge in smoke concentration is monitored. The time of abnormal change should be affected by the concentration, that is, the higher the concentration, the shorter the monitoring time. Therefore, the real-time smoke concentration data of the collection point can be continuously changed to obtain the real-time smoke surge index of the monitoring point. Taking a monitoring point as an example, the specific steps are as follows.
[0029] S2: Identify concentration anomalies based on concentration data and preset safety thresholds, and dynamically obtain monitoring time based on concentration anomalies, where concentration anomalies are inversely proportional to monitoring time. Extract concentration anomaly value sequences within the monitoring time, and calculate the smoke surge index to determine suspected fire sources.
[0030] Concentration outliers include: In response to the detected concentration data being greater than or equal to a preset safety threshold, the concentration abnormality value of the smoke is the difference between the concentration data and the preset safety threshold. Conversely, if the concentration data is less than the preset safety threshold, the concentration abnormality value is 0.
[0031] Specifically, the concentration anomaly value satisfies the following relationship: ; In the formula, Indicates The concentration abnormal value at the moment, Indicates monitoring point Time concentration data, Indicates the preset safety threshold.
[0032] In this embodiment, the preset safety threshold is: 0.15. By monitoring the deviation between the concentration data and the preset safety threshold, the potential fire risk is analyzed, so that a response can be made based on slight changes in the data, thereby improving the distinction between concentration data of normal smoke concentration and abnormal concentration data that may indicate a fire, and reducing false alarms.
[0033] The monitoring time satisfies the following relationship: ; In the formula, Indicates the monitoring time. Indicates The concentration abnormal value at the moment, Indicates the standard monitoring time, Represents a natural constant.
[0034] In this example, the standard monitoring time is 50.
[0035] That is to say, the time of abnormal change should be affected by the concentration, that is, the higher the concentration, the shorter the monitoring time, so as to dynamically adjust the time interval of the next monitoring. The system can dynamically adjust the monitoring frequency according to the actual monitored concentration data to achieve adaptive monitoring. Reducing the monitoring frequency when the smoke concentration is low can save resources, such as energy consumption and data processing capacity, while ensuring the monitoring effect.
[0036] In addition, in another embodiment, a fuzzy logic controller is used to determine the monitoring time, which involves converting input variables into fuzzy sets, reasoning through fuzzy rules, and then outputting a specific detection time, wherein the variables include but are not limited to: concentration data, concentration abnormality, temperature, humidity, etc. The following are the specific operation steps: Determine the input variables that affect the monitoring time, and define one or more fuzzy sets for each input variable, wherein each set represents a different state and condition, such as high, medium, and low.
[0037] Assign a membership function to the fuzzy set of each input variable, create a fuzzy rule base, perform reasoning based on the current input variable values and the fuzzy rule base, obtain the activation degree of each rule, and convert the fuzzy output into a specific detection time value.
[0038] To further illustrate, through the membership function, the specific physical quantity value is mapped to the membership value, which represents the degree to which the physical quantity belongs to a fuzzy set. The fuzzy rule base is based on expert knowledge or experience, describing the relationship between input variables and their impact on detection time.
[0039] Exemplarily, the rules in the fuzzy rule base are: if the concentration data is high, the concentration abnormality value is high, and the temperature is high, the monitoring time is short.
[0040] Further explanation: the result of fuzzy reasoning is converted into a specific monitoring time value, which is usually achieved by using a defuzzification method, such as the centroid method or the maximum membership principle. The centroid method calculates the weighted average of all possible detection time values, and the weight is the output membership value. Specifically, the centroid method is a well-known technology in the art and will not be described in detail.
[0041] The Smog Surge Index includes: Get the monitoring time window length , get the Time to The concentration anomaly value sequence at a certain moment is taken as the target moment. The ratio between the concentration anomaly value at the target moment and the concentration anomaly value at the monitoring time before the target moment is calculated. The standard deviation of the concentration anomaly value sequence is calculated to measure the degree of dispersion of each data point in the sequence relative to the mean value. The standard deviation is exponentially decayed, and the product of the ratio and the standard deviation after exponential decay is taken as the smoke surge index at the target moment.
[0042] The smoke surge index satisfies the following relationship: ; In the formula, Indicates The smoke surge index at the moment, Indicates The concentration abnormal value at the moment, Indicates The concentration abnormal value at the moment, Indicates the monitoring time length is The concentration outlier sequence of Indicates The exponential function with base , Represents the standard deviation function.
[0043] Further explanation: the concentration outlier sequence is ,in, Indicates The concentration abnormal value at the moment, Indicates The concentration abnormal value at the moment, Indicates The concentration abnormal value at the moment, Indicates The concentration outlier at the moment.
[0044] In addition, in another embodiment, the moving average change rate is a method for calculating the smoke surge index, which is achieved by comparing the current concentration abnormal value with the moving average of the concentration abnormal values in the past period of time. The operation steps are as follows: Get the monitoring time window length , calculate the past The moving average of the concentration anomaly values at each time point is taken as the target moment. The difference between the concentration anomaly value at the target moment and the moving average value is calculated, and the ratio of the difference to the moving average value is taken as the smoke surge index.
[0045] Specifically, the moving average satisfies the following relationship: ; In the formula, represents the moving average, Indicates the length of the monitoring time window, Indicates The concentration outlier at the moment.
[0046] That is, the moving average provides a smoother trend of concentration changes, helping to reduce the impact of short-term fluctuations and highlighting long-term trends.
[0047] Specifically, the smoke surge index satisfies the following relationship: ; In the formula, Indicates The smoke surge index at the moment, Indicates The concentration abnormal value at the moment, Represents a moving average.
[0048] It is further explained that the smoke surge index reflects the change in the current concentration anomaly relative to the average level. If the smoke surge index is large, it means that the current concentration anomaly is significantly higher than the average level, which may indicate a fire or a rapid increase in smoke concentration; if the smoke surge index is close to zero, it means that the current concentration anomaly is close to the average level and the smoke concentration has not changed significantly; if the smoke surge index is negative, it means that the current concentration anomaly is lower than the average level and the smoke concentration may be decreasing.
[0049] Further analysis shows that if the concentration abnormality is significantly higher than the average level due to smoke generated during cooking, although it is not a sign of fire, it indicates a significant increase in smoke concentration, which will then decrease with the ventilation system. If the smoke surge index decreases to near 0 or becomes negative during the subsequent monitoring period, it indicates that the smoke generated during cooking has been effectively removed by the ventilation system, or that the generation of cooking smoke is not continuous, resulting in the concentration abnormality not being significantly higher or even lower than the average level; in addition, bathroom steam and cigarette smoke can be removed by the ventilation system in a short period of time. Therefore, the change trend of the smoke surge index provides important information about the nature of the smoke source and the efficiency of the ventilation system. If the smoke surge index continues to remain at a high level or continues to increase, this may indicate the presence of a persistent smoke source that requires further investigation. On the contrary, if the smoke surge index decreases to normal levels or below, this indicates that the smoke source has been controlled or eliminated, and it can more accurately distinguish between smoke generated by daily activities and abnormal smoke that may indicate a fire. This approach helps reduce false alarms while ensuring a rapid response to real fires.
[0050] At this point, all monitoring points are traversed to obtain the real-time smoke surge index corresponding to each monitoring point.
[0051] It should be noted that the smoke surge index corresponding to each monitoring point alone cannot be used to directly determine whether a fire has occurred, because non-fire conditions such as cooking smoke and cigarette smoke can also cause changes in the smoke surge index of a certain monitoring point. Therefore, it is necessary to determine whether the monitoring point is a real fire monitoring point.
[0052] That is to say, the closer the monitoring point is to the fire point, the higher the smoke surge index is, and the smoke surge index of the monitoring points around the nearest monitoring point decreases as the distance increases. That is, the smoke surge index of the monitoring points around the fire point corresponding to the distance to the fire point shows a gradient decrease trend, and the gradient decrease is not linear due to the influence of ventilation, that is, the gradient of the smoke surge index of the monitoring points at the same distance from the fire point may be inconsistent. Therefore, the continuous value of the gradient of the smoke surge index of the monitoring point is calculated as the fire point. The specific steps are as follows.
[0053] S3: Calculate the continuity index of the smoke surge index of the monitoring point, multiply the product of the continuity index at all times as the continuity degree of the monitoring point, determine the fire point according to the continuity degree, and take the minimum value point of the smoke surge index as the monitoring starting point.
[0054] Calculating the continuous index of the smoke surge index of the monitoring point includes the steps of: Take any monitoring point as the target monitoring point and obtain the smoke surge index at each moment; In response to Moment Smoke surge index at each monitoring point Greater than Moment ago Smoke surge index at each monitoring point , then the continuity index is 1; In response to Moment Smoke surge index at each monitoring point Less than or equal to Smoke surge index at each monitoring point , the continuity index is 0.
[0055] Specifically, the continuity index satisfies the following relationship: ; In the formula, Indicates Moment The continuous index of each monitoring point, Indicates Moment The smoke surge index at each monitoring point is Indicates Moment The smoke surge index at each monitoring point.
[0056] The degree of continuity satisfies the following relationship: ; In the formula, Indicates The degree of continuity of time, represents the total number of monitoring points, Indicates Moment Continuous index of each monitoring point; In response to the continuity degree being 1, the monitoring points are continuous and in one smoke diffusion channel; conversely, when the continuity degree is 0, the monitoring points are not on the smoke diffusion channel, and multiple diffusion channels can be obtained, each of which includes multiple monitoring points.
[0057] Further explanation, by analyzing the continuity of the smoke surge index, true fire events can be more accurately identified and distinguished from smoke surges due to non-fire reasons such as cooking, bathroom steam or cigarette smoke. The degree of continuity helps reduce false alarms because multiple monitoring points are required to consistently show smoke surges, which are more likely to be signs of fire rather than random or localized smoke events, and by identifying smoke dispersion channels, the source of the fire can be located more quickly, thereby speeding up emergency response time. Understanding smoke dispersion channels can help optimize the allocation of firefighting resources and ensure that firefighting forces are quickly deployed where they are most needed.
[0058] S4: Predict the monitoring starting point based on the preset prediction network, obtain the probability of the smoke dispersion point, predict the distribution of smoke at future time points, and issue a warning alarm based on the prediction results.
[0059] Labels are set for the monitoring points on the diffusion channel, and the monitoring point matrix is obtained according to the positions of the monitoring points; To further illustrate, the monitoring points on the diffuse channel are marked as 1, and the monitoring points on the non-diffuse channel are marked as 0.
[0060] Smoke points are artificially created at preset positions, and a preset number of real-time monitoring matrices of smoke points in a diffusion process are collected as a training set for a training, and prediction matrices at multiple moments are obtained. The prediction matrices are put into a preset prediction network, and in response to the loss function being less than a preset value or reaching a preset number of training times, a smoke diffusion point prediction network is obtained.
[0061] It should be noted that, in the process of collecting smoke point diffusion once, illustratively, 100 monitoring points are set, and the monitoring point matrix of the 100 monitoring points at multiple times is obtained as a training set.
[0062] The smoke dispersion point prediction network is a convolutional neural network, including: input layer: receiving raw input data; convolution layer: applying learnable filters to slide on the input data to capture local features; pooling layer: used to reduce the dimension of feature space and reduce the amount of calculation; fully connected layer: extracting features and mapping them to the final output; output layer: outputting category probabilities through activation functions.
[0063] The real-time monitoring matrix is put into the smoke diffusion point prediction network to obtain the prediction matrix for the next moment in the future. The monitoring point corresponding to 1 point in the prediction matrix is the smoke diffusion point. When the smoke diffusion point is greater than the preset value, a fire alarm is issued.
[0064] In this embodiment, the preset value is 0.5.
[0065] The present invention also provides a smart city fire monitoring and early warning system based on the Internet of Things. Figure 3As shown, the system includes a processor and a memory, and the memory stores computer program instructions. When the computer program instructions are executed by the processor, the smart city fire monitoring and early warning method based on the Internet of Things according to the first aspect of the present invention is implemented.
[0066] The system also includes other components familiar to those skilled in the art, such as a communication bus and a communication interface. The configuration and functions of these components are known in the art and will not be described in detail here.
[0067] In the present invention, the aforementioned memory may be any tangible medium containing or storing a program that can be used by or in combination with an instruction execution system, apparatus or device. For example, a computer-readable storage medium may be any appropriate magnetic storage medium or magneto-optical storage medium, such as a resistive random access memory RRAM (Resistive Random Access Memory), a dynamic random access memory DRAM (Dynamic Random Access Memory), a static random access memory SRAM (Static Random-Access Memory), an enhanced dynamic random access memory EDRAM (Enhanced Dynamic Random Access Memory), a high-bandwidth memory HBM (High-Bandwidth Memory), a hybrid memory cube HMC (Hybrid Memory Cube), etc., or any other medium that can be used to store the required information and can be accessed by an application, a module, or both. Any such computer storage medium may be part of a device or accessible or connectable to a device. Any application or module described in the present invention may be implemented using computer-readable / executable instructions that may be stored or otherwise maintained by such a computer-readable medium.
[0068] In the description of this specification, "plurality" or "several" means at least two, such as two, three or more, etc., unless otherwise clearly and specifically defined.
[0069] Although this specification has shown and described a number of embodiments of the present invention, it will be apparent to those skilled in the art that such embodiments are provided by way of example only. Those skilled in the art will conceive of many modifications, changes and alternatives without departing from the ideas and spirit of the present invention. It should be understood that in the practice of the present invention, various alternatives to the embodiments of the present invention described herein may be employed.
Claims
1. A fire monitoring and early warning method for a smart city based on the Internet of Things, characterized in that: include: Obtain smoke concentration data at each monitoring point in real time; Identify concentration abnormal values according to concentration data and preset safety thresholds, dynamically obtain monitoring time based on the concentration abnormal values, wherein the concentration abnormal values are inversely proportional to the monitoring time, extract concentration abnormal value sequences within the monitoring time, and calculate the smoke surge index to determine suspected fire sources; Calculate the continuous index of the smoke surge index of the monitoring point, multiply the continuous index product of all moments as the continuity degree of the monitoring point, determine the fire point according to the continuity degree, and take the minimum value point of the smoke surge index as the monitoring starting point; Predicting the monitoring starting point based on a preset prediction network, obtaining the probability of the smoke dispersion point, predicting the distribution of smoke at a future time point, and issuing a warning alarm based on the prediction result; Wherein, the continuity index satisfies the following relationship: ; In the formula, Indicates Moment The continuous index of each monitoring point, Indicates Moment The smoke surge index at each monitoring point is Indicates Moment The smoke surge index at each monitoring point.
2. The method for fire monitoring and early warning in smart cities based on the Internet of Things according to claim 1 is characterized in that: The abnormal concentration values include: In response to the detected concentration data being greater than or equal to a preset safety threshold, the concentration abnormality value of the smoke is the difference between the concentration data and the preset safety threshold. Conversely, if the concentration data is less than the preset safety threshold, the concentration abnormality value is 0.
3. The method for fire monitoring and early warning in smart cities based on the Internet of Things according to claim 1 is characterized in that: The monitoring time satisfies the following relationship: ; In the formula, Indicates the monitoring time. Indicates The concentration abnormal value at the moment, Indicates the standard monitoring time, Represents a natural constant.
4. The method for fire monitoring and early warning in smart cities based on the Internet of Things according to claim 1 is characterized in that: The Smog Surge Index includes: Get the monitoring time window length , get the Time to The concentration anomaly value sequence at a certain moment is taken as the target moment. The ratio between the concentration anomaly value at the target moment and the moment of monitoring time before the target moment is calculated. The standard deviation of the concentration anomaly value sequence is calculated to measure the degree of dispersion of each data point in the sequence relative to the mean value. The standard deviation is exponentially decayed, and the product of the ratio and the exponentially decayed standard deviation is taken as the smoke surge index at the target moment.
5. The method for fire monitoring and early warning in smart cities based on the Internet of Things according to claim 1 is characterized in that: The degree of continuity includes: In response to the continuity degree being 1, the monitoring points are continuous and in a smoke diffusion channel; conversely, when the continuity degree is 0, the monitoring points are not on the smoke diffusion channel, and multiple diffusion channels can be obtained, each diffusion channel includes multiple monitoring points, labels are set for the monitoring points on the diffusion channels, and the monitoring point matrix is obtained by arranging the monitoring point positions.
6. The method for fire monitoring and early warning in smart cities based on the Internet of Things according to claim 1 is characterized in that: The predicting the monitoring starting point based on a preset prediction network includes: Smoke points are artificially created at preset positions, and a preset number of real-time monitoring matrices of smoke points in a diffusion process are collected as a training set for a training, and prediction matrices at multiple moments are obtained. The prediction matrices are put into a preset prediction network, and in response to a loss function being less than a preset value or reaching a preset number of training times, a smoke diffusion point prediction network is obtained.
7. The method for fire monitoring and early warning in a smart city based on the Internet of Things according to claim 6 is characterized in that: The smoke dispersion point prediction network is a convolutional neural network. Includes: Input layer: receives raw input data; Convolutional layer: applies a learnable filter that slides over the input data to capture local features; Pooling layer: used to reduce the dimension of feature space and reduce the amount of calculation; Fully connected layer: extracts features and maps them to the final output; Output layer: outputs category probability through activation function; The real-time monitoring matrix is put into the smoke diffusion point prediction network to obtain the prediction matrix for the next moment in the future. The monitoring point corresponding to 1 point in the prediction matrix is the smoke diffusion point. When the smoke diffusion point is greater than the preset value, a fire alarm is issued.
8. The smart city fire monitoring and early warning system based on the Internet of Things is characterized by: include: A processor and a memory, wherein the memory stores computer program instructions, and when the computer program instructions are executed by the processor, the smart city fire monitoring and early warning method based on the Internet of Things according to any one of claims 1 to 7 is implemented.
Citation Information
Cited By
Intelligent smoke monitoring data processing method and device based on 4G network
CN119380474A
Fire smoke particle detection method and system for intelligent fire fighting
CN120220314A
