Fire safety early warning method and system based on wireless networking

Abstract

The application discloses a fire safety early warning method and system based on wireless networking, and relates to the technical field of fire safety early warning.The fire safety early warning method based on wireless networking obtains power and environmental monitoring data of multiple areas in a wireless network of a shopping mall in real time, analyzes electrical fire hazard inducing indexes and fire hazard environmental disturbance indexes;meanwhile, obtains shopping mall monitoring image angle sequence data of each area, analyzes visible safety hidden danger indexes based on a pre-trained fire safety identification model, and comprehensively evaluates the safety conditions of each area together with the electrical fire hazard inducing indexes and the fire hazard environmental disturbance indexes, to finally generate comprehensive safety perception indexes of each area.The application performs fire safety early warning on the corresponding area of each area through the comprehensive safety perception indexes of each area, thereby identifying and locating fire hazards in a multi-area and complex layout environment such as a shopping mall in real time, and effectively improving the accuracy of fire safety early warning, and then improving the overall fire safety level.

Description

Technical Field

[0001] This invention relates to the field of fire safety early warning technology, specifically to a fire safety early warning method and system based on wireless networking. Background Technology

[0002] Wireless networking technology is a network technology that uses wireless communication to connect multiple node devices in a self-organizing manner. It has advantages such as flexible networking, wide coverage, and strong self-healing ability. With the rapid development of sensor technology and wireless communication technology, wireless networking provides strong technical support for real-time monitoring in large-scale and dynamic environments. For example, in public places such as large shopping malls, fire safety management has complex and diverse characteristics. The mall is densely populated, fire-fighting facilities are widely distributed, electrical equipment is complex, and environmental conditions are variable, which can easily lead to electrical fire hazards, environmental fire hazards, and equipment malfunctions. Traditional fire safety monitoring methods often rely on wired systems, which are costly to install and lack flexibility, making it difficult to meet the needs of real-time, dynamic, and full-coverage monitoring.

[0003] Existing technology, such as the fire early warning method and system disclosed in patent application CN111882800B, includes the following steps: when abnormal lighting and / or abnormal sound and / or abnormal force related to fire are present, the system determines whether there is a potential fire safety hazard based on temperature sensor data and / or smoke concentration alarm data and / or camera data; if so, a fire early warning is issued. The method and system of this invention solve the technical problem of how to combine multi-dimensional data such as sound, light, and force for fire detection and early warning.

[0004] Based on the above findings, the limitations of existing technologies include at least the following problems: Existing technologies fail to effectively integrate multi-dimensional real-time data, making it difficult to cope with changes in multi-source data in modern complex environments, especially in dynamic commercial environments such as shopping malls. Environmental parameters and the operating status of electrical equipment are changing in real time and there are mutual interference relationships between them. For example, the use of large electrical equipment such as air conditioners and elevators will increase the load on the electrical equipment, causing the temperature to rise. Such instantaneous changes in load and temperature rise are often precursors to fire hazards. Existing technologies ignore the inherent connection between the two, making it difficult for existing technologies to identify potential fire risks in complex environments, thereby affecting the accuracy of fire safety early warning. Summary of the Invention

[0005] To address the shortcomings of existing technologies, this invention provides a fire safety early warning method and system based on wireless networking, which solves the problem that existing technologies fail to effectively integrate multi-dimensional real-time data, resulting in inaccurate fire risk identification in complex environments.

[0006] To achieve the above objectives, the present invention provides the following technical solution: a fire safety early warning method based on wireless networking, comprising the following steps: real-time acquisition of power monitoring data and environmental monitoring data of several areas within a designated shopping mall wireless network, and analysis of the electrical fire hazard induction index and fire environmental disturbance index of the corresponding areas; acquisition of shopping mall monitoring image angle sequence data for each area within the designated shopping mall wireless network, and comprehensive analysis based on a pre-trained fire safety identification model to obtain the visible safety hazard index of the corresponding area; analysis of the comprehensive safety perception index of the corresponding area based on the electrical fire hazard induction index, fire environmental disturbance index, and visible safety hazard index of each area within the designated shopping mall wireless network; and fire safety early warning for the corresponding area based on the comprehensive safety perception index of each area within the designated shopping mall wireless network.

[0007] Furthermore, the power monitoring data includes power load thermal index, cable leakage microcurrent value, contact resistance drift value, load connection surge intensity value, spike pulse interference intensity value, and arc disturbance value. The specific steps for analyzing and setting the electrical fire hazard induction index for each area within the shopping mall wireless network are as follows: Based on the power monitoring data for each area within the shopping mall wireless network, analyze the electrical fire hazard assessment set for that area, including the electrical overload thermal change index and the electrical transient anomaly index; based on the electrical fire hazard assessment set for each area within the shopping mall wireless network, analyze the electrical fire hazard induction index for that area.

[0008] Furthermore, the specific steps for analyzing and setting up the electrical fire hazard assessment set for each area within the shopping mall wireless network are as follows: Based on Z-score standardization, the power load thermal index, cable leakage micro-current value, and contact resistance drift value of each area within the shopping mall wireless network are weighted to obtain the electrical overload thermal index of the corresponding area; Based on Z-score standardization, the load connection surge intensity value, spike pulse interference intensity value, and arc disturbance value of each area within the shopping mall wireless network are weighted to obtain the electrical transient anomaly index of the corresponding area.

[0009] Furthermore, the environmental monitoring data includes combustible gas disturbance index, air temperature value, smoke concentration value, air ion concentration fluctuation index, and infrared blocking disturbance rate value. The specific steps for analyzing and setting the fire hazard environmental disturbance index of each area within the shopping mall wireless network are as follows: Based on the environmental monitoring data of each area within the shopping mall wireless network, analyze the environmental fire hazard assessment set of its corresponding area, including combustible factor anomaly index and thermal radiation anomaly index; based on the environmental fire hazard assessment set of each area within the shopping mall wireless network, analyze the fire hazard environmental disturbance index of its corresponding area.

[0010] Furthermore, the specific steps for analyzing and setting up the environmental fire hazard assessment set for each area within the shopping mall's wireless network are as follows: Based on Z-score standardization, the combustible gas disturbance index and air ion concentration fluctuation index of each area within the shopping mall's wireless network are weighted to obtain the corresponding combustible factor anomaly index; Based on Z-score standardization, the air temperature value, smoke concentration value, and infrared blocking disturbance rate value of each area within the shopping mall's wireless network are weighted to obtain the corresponding combustible factor anomaly index.

[0011] Furthermore, the shopping mall surveillance image angle sequence data specifically refers to the shopping mall surveillance image data for each angle, all of which include the pixel value and two-dimensional coordinates of each pixel. The specific steps for analyzing and setting the visual safety hazard index of each area within the shopping mall wireless network are as follows: input the shopping mall surveillance image data of each area within the shopping mall wireless network into a pre-trained fire safety recognition model for comprehensive analysis to obtain the visual hazard assessment set of the corresponding area, including the visual thermal disturbance perception index and the fire protection component deterioration index; based on the visual hazard assessment set of each area within the shopping mall wireless network, analyze the visual safety hazard index of the corresponding area.

[0012] Furthermore, the specific formula for calculating the visible security hazard index of a certain area within the shopping mall's wireless network is as follows: Among them, KaQ is the visual safety hazard index of a certain area in the shopping mall wireless network, SrD is the visual thermal interference perception index of a certain area in the shopping mall wireless network, α1 is the thermal interference coefficient stored in the database, XgJ is the fire protection component deterioration index of a certain area in the shopping mall wireless network, α2 is the deterioration coefficient stored in the database, and α3 is the interaction coefficient stored in the database.

[0013] Furthermore, the fire safety identification model includes a feature extraction subnetwork, a feature fusion subnetwork, and an evaluation output subnetwork. The specific steps for obtaining the visual hazard assessment set for each area within the designated shopping mall wireless network are as follows: In the feature extraction subnetwork of the fire safety identification model, shopping mall monitoring image data from each angle of each area within the designated shopping mall wireless network are received and extracted to obtain the feature vector set for the corresponding angle; In the feature fusion subnetwork of the fire safety identification model, the feature vector set for each angle of each area within the designated shopping mall wireless network is classified and fused to obtain the joint feature vector set for the corresponding area; In the evaluation output subnetwork of the fire safety identification model, the joint feature vector set for each area within the designated shopping mall wireless network is mapped to obtain the visual hazard assessment set for the corresponding area.

[0014] Furthermore, the specific formula for calculating the comprehensive security perception index of a certain area within the shopping mall's wireless network is as follows: Wherein, ZaG is the comprehensive security perception index of a certain area within the shopping mall wireless network, DqH is the electrical fire hazard induction index of a certain area within the shopping mall wireless network, β1 is the electrical coefficient stored in the database, HxJ is the fire environmental disturbance index of a certain area within the shopping mall wireless network, β2 is the environmental coefficient stored in the database, KaQ is the visible safety hazard index of a certain area within the shopping mall wireless network, β3 is the visibility coefficient stored in the database, β4 is the electrical environment coordination coefficient stored in the database, and ω is the adjustment coefficient stored in the database.

[0015] The fire safety early warning system based on wireless networking includes: an electrical fire hazard analysis module, used to acquire real-time power monitoring data and environmental monitoring data of several areas within the designated shopping mall wireless network, and analyze the electrical fire hazard induction index of the corresponding areas; an environmental fire hazard analysis module, used to acquire environmental monitoring data of each area within the designated shopping mall wireless network, and analyze the fire environmental disturbance index of the corresponding area; a visual safety hazard analysis module, used to acquire the angle sequence data of shopping mall monitoring images of each area within the designated shopping mall wireless network, and combine it with a pre-trained fire safety identification model for comprehensive analysis to obtain the visual safety hazard index of the corresponding area; a comprehensive safety perception early warning module, used to analyze the comprehensive safety perception index of the corresponding area based on the electrical fire hazard induction index, fire environmental disturbance index, and visual safety hazard index of each area within the designated shopping mall wireless network; and a fire safety early warning feedback module, used to provide fire safety early warnings for the corresponding area based on the comprehensive safety perception index of each area within the designated shopping mall wireless network.

[0016] The present invention has the following beneficial effects:

[0017] (1) The fire safety early warning method based on wireless networking collects multiple monitoring data from multiple areas in the shopping mall in real time at high frequency through wireless networking, analyzes and processes them separately, generates corresponding indices, and performs fusion early warning on this basis to obtain the comprehensive safety perception index of the corresponding area. It not only reflects the risk status of a single data source, but also reflects the interaction and superposition effect between various risk factors, so as to discover potential fire hazards more quickly and accurately, and thereby identify and locate fire hazards in multiple areas and complex layout environments such as shopping malls in real time, thereby effectively improving the accuracy of fire safety early warning and improving the overall fire safety level.

[0018] (2) The fire safety early warning method based on wireless networking acquires the angle sequence data of the shopping mall monitoring images in real time, and extracts the visual hazard related features through the pre-trained fire safety identification model, thereby generating the visual safety hazard index of each area. This index measures the visual safety status and potential risks of fire protection facilities and environment in the area. It can also effectively capture safety hazards with subtle changes, thereby comprehensively improving the accuracy of fire hazard identification in complex environments. It can also discover some hidden dangers that are not easily detected in advance, thereby helping relevant personnel to grasp the fire safety status in real time and significantly enhancing the fire prevention and control capabilities and safety assurance level of shopping malls and similar complex spaces.

[0019] (3) The fire safety early warning method based on wireless networking comprehensively analyzes power monitoring data and environmental monitoring data, and uses the entropy weight method to accurately analyze the weighting of various monitoring indicators. It also combines Z-score standardization technology to normalize the data, thereby effectively improving the accuracy of the fusion between monitoring indicators and generating corresponding indices. It comprehensively reflects the mutual influence of different monitoring indicators, thereby achieving a comprehensive assessment of the potential fire risks in different areas of the shopping mall, improving the flexibility of the shopping mall's fire emergency response, and effectively reducing the probability of fire accidents, thus ensuring the safe operation of the shopping mall.

[0020] (4) The fire safety early warning system based on wireless networking integrates multiple modules to achieve comprehensive monitoring and real-time data analysis of multiple areas in the shopping mall. Each module monitors and analyzes the corresponding monitoring data to accurately identify the fire risk characteristics of each area. Especially in the complex environment of the shopping mall, the system can classify and process different types of hidden dangers and generate a comprehensive safety perception index for each area in combination with the comprehensive safety perception early warning module. This provides managers with accurate fire early warning information, enabling the shopping mall to achieve accurate early warning and location before a fire occurs, thereby improving the shopping mall's fire prevention and control capabilities.

[0021] Of course, any product implementing this invention does not necessarily need to achieve all of the advantages described above at the same time. Attached Figure Description

[0022] Figure 1 This is a flowchart of the fire safety early warning method based on wireless networking according to the present invention.

[0023] Figure 2 This is a flowchart illustrating the specific steps involved in analyzing and setting the visible safety hazard index for each area within a shopping mall's wireless network, as part of the fire safety early warning method based on wireless networking, according to the present invention.

[0024] Figure 3This is a schematic diagram illustrating the sequence of visual thermal disturbance perception index regions within a shopping mall wireless network in the fire safety early warning method based on wireless networking of the present invention.

[0025] Figure 4 This is a schematic diagram illustrating the sequence of fire safety component deterioration index areas within a shopping mall's wireless network, as defined in the fire safety early warning method based on wireless networking of this invention.

[0026] Figure 5 This is a block diagram of the fire safety early warning system based on wireless networking according to the present invention. Detailed Implementation

[0027] Please see Figure 1 This invention provides a technical solution: a fire safety early warning method based on wireless networking, comprising the following steps: real-time acquisition of power monitoring data and environmental monitoring data of several (monitoring) areas within a designated shopping mall wireless network (multiple sensors and monitoring devices are deployed in multiple areas within a designated shopping mall area to monitor and transmit data in real time, using a mesh topology, such as Mesh, with the area as the main node and multiple sensors and monitoring devices within the area as child nodes, each sensor monitoring data related to fire safety in the area, each child node transmitting data to the main node of the area via wireless communication, and each main node also transmitting data via wireless communication, thus forming a shopping mall wireless network); analysis of the electrical fire hazard induction index and fire environment disturbance index of the corresponding area; acquisition of shopping mall monitoring image angle sequence data of each area within the designated shopping mall wireless network, and comprehensive analysis combined with a pre-trained fire safety identification model to obtain the visible safety hazard index of the corresponding area; analysis of the comprehensive safety perception index of the corresponding area based on the electrical fire hazard induction index, fire environment disturbance index, and visible safety hazard index of each area within the designated shopping mall wireless network;

[0028] Based on the comprehensive security perception index of each area within the shopping mall's wireless network, fire safety warnings are issued for the corresponding areas. Specifically, the process involves: determining whether the comprehensive security perception index of each area within the shopping mall's wireless network is higher than a preset comprehensive security perception index threshold; if it is not higher than the preset threshold, no fire safety warning is issued for the corresponding area; if it is higher than the preset threshold, a fire safety warning is issued for the corresponding area, and the safety warning level of the corresponding area is analyzed, that is, comparing the comprehensive security perception index of the area requiring a fire safety warning with a preset risk threshold range; if it is lower than the lower limit of the preset risk threshold range, a green warning (i.e., low risk) is issued; if it is within the preset risk threshold range, a yellow warning (i.e., medium risk) is issued; and if it is higher than the upper limit of the preset risk threshold range, a red warning (i.e., high risk) is issued.

[0029] The specific formula for calculating the comprehensive security perception index of a certain area within the shopping mall's wireless network is as follows: Wherein, ZaG is the comprehensive safety perception index of a certain area within the shopping mall wireless network, DqH is the electrical fire hazard induction index of a certain area within the shopping mall wireless network, β1 is the electrical coefficient stored in the database, HxJ is the fire environmental disturbance index of a certain area within the shopping mall wireless network, β2 is the environmental coefficient stored in the database, KaQ is the visible safety hazard index of a certain area within the shopping mall wireless network, β3 is the visibility coefficient stored in the database, β4 is the electrical environment coordination coefficient stored in the database, and ω is the adjustment coefficient stored in the database, which is taken as 3.000 in this implementation example.

[0030] It should be explained that the steps for obtaining the electrical coefficient β1 stored in the database are as follows: obtain the historical electrical fire hazard indices at several historical time points (with the same logic as obtaining the electrical fire hazard indices), and analyze the mean and standard deviation of the historical electrical fire hazard indices respectively, and perform ratio processing, that is, the standard deviation of the historical electrical fire hazard indices / the mean of the historical electrical fire hazard indices, and use the ratio processing result as the electrical coefficient β1.

[0031] The steps for obtaining the environmental coefficient β2 stored in the database are as follows: Obtain the historical fire risk environmental disturbance index at several historical time points (with the same logic as obtaining the fire risk environmental disturbance index), and analyze the mean and standard deviation of the historical fire risk environmental disturbance index respectively, and perform ratio processing, that is, the standard deviation of the historical fire risk environmental disturbance index / the mean of the historical fire risk environmental disturbance index, and use the ratio processing result as the environmental coefficient β2.

[0032] The steps for obtaining the visibility coefficient β3 stored in the database are as follows: Obtain the historical fire hazard environment disturbance index at several historical time points (with the same logic as obtaining the fire hazard environment disturbance index), and analyze the mean and standard deviation of the historical visible safety hazard index respectively, and perform ratio processing, that is, the standard deviation of the historical visible safety hazard index / the mean of the historical visible safety hazard index, and use the ratio processing result as the visibility coefficient β3.

[0033] The steps for obtaining the electrical environment coordination coefficient β4 stored in the database are as follows: Read the historical electrical fire hazard induction index and historical fire environment disturbance index for several historical time points in the region, and extract the historical coordination sum and historical coordination interaction value (i.e., electrical fire hazard induction index × fire environment disturbance index) respectively. Then, the historical electrical fire hazard induction index and historical fire environment disturbance index at each historical time point are compared with the historical coordination sum. The ratio processing result is weighted with the historical coordination interaction value, and the mean is calculated based on the result. The result is then used as the electrical environment coordination coefficient β4.

[0034] Specifically, the power monitoring data includes power load thermal index, cable leakage microcurrent value, contact resistance drift value, load connection surge intensity value, spike pulse interference intensity value, and arc disturbance value. The specific steps for analyzing and setting the electrical fire hazard induction index for each area within the shopping mall wireless network are as follows: Based on the power monitoring data for each area within the shopping mall wireless network, analyze the corresponding area's electrical fire hazard assessment set, including the electrical overload thermal change index (characterizing the risk intensity of continuous heating caused by electrical line overload, aging, insulation deterioration, loose connection, etc.) and the electrical transient anomaly index (characterizing the risk intensity of abnormal electrical behavior).

[0035] Based on the electrical fire hazard assessment set for each area within the shopping mall's wireless network, the electrical fire hazard induction index for each area is analyzed. This involves weighting the electrical overload thermal distortion index and electrical transient anomaly index for each area; the result is the corresponding electrical fire hazard induction index. It should be noted that the weighting coefficients for each parameter in this implementation plan are obtained using the entropy weighting method. Taking the acquisition of the weighting coefficients for the electrical overload thermal distortion index and electrical transient anomaly index as an example, for each monitoring area, the electrical overload thermal distortion index and electrical transient anomaly index are collected within the same time window and formed into a multi-row, multi-column data table. Each row represents sample data from one monitoring area, and each column represents an independent... The parameters are calculated and proportionalized by dividing the value of a certain indicator in each monitoring area by the sum of the values ​​of that indicator in all areas. This calculates the proportion of that value in the total number of indicators in that area. This proportion represents the relative distribution of that indicator in the overall sample. Then, based on the information entropy theory, the uniformity of the distribution of each type of indicator in all areas is analyzed. After calculating the entropy value of each type of indicator, the redundancy of each indicator is calculated (i.e., 1 minus the normalized entropy value of that indicator). This redundancy is then proportionalized to the sum of the redundancy of all indicators. The resulting proportion is the effective information proportion of that indicator. The sum of the redundancy of all indicators is used as the normalization reference value. Finally, the ratio between the redundancy of each indicator and the sum is used as its corresponding weighting coefficient.

[0036] The power load thermal index represents the risk of current-carrying heat generation for all lines in the area. It is obtained by acquiring the current and voltage values ​​of each line using current and voltage sensors, and then comparing them with preset reference current values ​​(obtained by averaging historical current values ​​from several time points) and reference voltage values ​​(obtained using the same logic as the reference current value). The ratios are then calculated as (current value / reference current value), and a weighted average is performed based on the ratio results to obtain the power load thermal index.

[0037] The cable leakage microcurrent value is the microcurrent leakage value of the cable trunk and cable branch lines to ground in this area (the microcurrent leakage phenomenon between cable conductors, such as live wires and neutral wires, and ground caused by insulation aging, moisture absorption, mechanical damage, or thermal carbonization). It can be obtained by using a high-sensitivity leakage current sensor to obtain the microcurrent leakage value of each cable trunk and cable branch line in this area, and then performing weighted average processing. The result is the cable leakage microcurrent value.

[0038] The contact resistance drift value is the stability of the contact state of electrical connection points (such as distribution terminals, socket connectors, etc.) in this area. It can be obtained by using current and voltage sensors installed at each electrical equipment contact point (such as wiring terminals, socket connectors, etc.) to obtain the current and voltage values ​​passing through the contact points, and by calculating the contact resistance value of each electrical equipment contact point using Ohm's law. The value is then compared with a preset corresponding contact resistance reference value (with the same logic as the reference current value acquisition) (contact resistance value / contact resistance reference value). The weighted average of the ratio results is then used to obtain the contact resistance drift value.

[0039] The load surge intensity value is the peak current impact caused by large electrical equipment (such as air conditioner compressors, elevator drives, etc.) in the area at the moment of power-on start-up. It is obtained by acquiring the peak current of each large electrical equipment through current sensors and performing weighted average processing. The result is the load surge intensity value.

[0040] The spike pulse interference intensity value is the amplitude of the spike voltage change signal in the power lines within the area. It is obtained by using EMI interference detectors deployed at each sub-line node (there are multiple sub-line nodes in the power lines within the area, which are connected to different types of electrical equipment in the area, such as lighting, elevators, refrigeration, advertising screens, etc.). The instantaneous voltage fluctuation intensity value of each sub-line node is obtained (that is, the instantaneous voltage value of each sub-line node obtained by the EMI interference detector exceeds the preset stable operating voltage value of 220V. If it exceeds, it is marked as the corresponding instantaneous voltage fluctuation intensity value; if not, it is directly 0). The result is obtained by weighted averaging.

[0041] The arc disturbance value is the disturbance intensity when an arc without open flame occurs at key points such as joints, cable trays, and electrical equipment interfaces in the region. It is obtained through an arc detection sensor, which is based on the principle of high-frequency electromagnetic wave changes and monitors the instantaneous disturbance signal generated during the arc discharge process in real time. The sensor can capture the sharp fluctuation of the electromagnetic wave amplitude when the arc occurs and perform weighted average processing on the disturbance intensity of each key node. The result is the arc disturbance value.

[0042] The specific steps for analyzing the electrical fire hazard assessment set of each area within the shopping mall's wireless network are as follows: Based on Z-score standardization, the power load thermal index, cable leakage microcurrent value, and contact resistance drift value of each area within the shopping mall's wireless network are weighted (i.e., Z-score standardization is performed on the corresponding area's power load thermal index, cable leakage microcurrent value, and contact resistance drift value, and then weighted based on the result) to obtain the electrical overload thermal index of the corresponding area; Based on Z-score standardization, the load surge intensity value, spike pulse interference intensity value, and arc disturbance value of each area within the shopping mall's wireless network are weighted (i.e., Z-score standardization is performed on the corresponding area's load surge intensity value, spike pulse interference intensity value, and arc disturbance value, and then weighted based on the result) to obtain the electrical transient anomaly index of the corresponding area.

[0043] In this implementation plan, by introducing Z-score standardization and entropy weighting, the problems of inconsistent dimensions and large numerical differences among various power monitoring indicators are effectively solved. Furthermore, it significantly enhances the ability to effectively highlight key risk characteristics in the overall assessment, thereby improving the reliability of the assessment results. Secondly, Z-score standardization unifies indicators with different physical units to a comparable standard scale, ensuring the comparability of various monitoring data under a unified standard. This helps to better align the subsequent weight allocation with the actual risk distribution. Meanwhile, the entropy weighting method automatically allocates weights based on the actual data distribution, highlighting which parameters play a dominant role in risk assessment, thus improving the accuracy of electrical fire hazard assessment. The resulting electrical overload thermal distortion index and electrical transient anomaly index can comprehensively quantify fire-inducing factors in electrical systems, thereby improving the reliability of fire safety early warning.

[0044] Specifically, the environmental monitoring data includes combustible gas disturbance index, air temperature value (obtained through a temperature sensor), smoke concentration value (obtained through a light scattering smoke sensor), air ion concentration fluctuation index, and infrared blocking disturbance rate value. The specific steps for analyzing and setting the fire hazard environmental disturbance index for each area within the shopping mall's wireless network are as follows: Based on the environmental monitoring data for each area within the shopping mall's wireless network, analyze the corresponding area's environmental fire hazard assessment set, including combustible factor anomaly index and thermal radiation anomaly index; Based on the environmental fire hazard assessment set for each area within the shopping mall's wireless network, analyze the corresponding area's fire hazard environmental disturbance index (i.e., weight the combustible factor anomaly index and thermal radiation anomaly index for each area, and the result is the corresponding fire hazard environmental disturbance index).

[0045] The combustible gas disturbance index is the risk of combustible gas leakage in the area. It is obtained by acquiring the concentration values ​​of each combustible gas (including but not limited to propane, butane, and carbon monoxide) through multiple gas sensors deployed in the area, and then processing the standard deviation to obtain the standard deviation value of the concentration of each combustible gas. The result is then weighted and processed to obtain the combustible gas disturbance index.

[0046] The air ion concentration fluctuation index reflects the instability of the charge state in the air in a given area. It is related to the initial stage of an electrical fire, which involves arc discharge without open flame and high-temperature gas excitation. It can instantly reflect abnormal electrical discharges and potential combustion hazards. Furthermore, by deploying multiple high-sensitivity air ion concentration sensors in the area, the instantaneous concentrations of positive and negative ions in the air are measured in real time based on the principle of electric field induction. These concentrations are then compared with preset positive ion reference concentrations (which are obtained using the same logic as the reference current value) and negative ion reference concentrations (which are obtained using the same logic as the reference current value) (instantaneous positive ion concentration / positive ion reference concentration). The weighted average of these ratios is then used to obtain the thermal index of the electrical load.

[0047] The infrared blocking perturbation rate is the degree of abnormal change in infrared signal caused by the enhancement of infrared radiation generated by flames or the obstruction of smoke in the area. It reflects the dynamic changes of flame generation, smoke obstruction and abnormal heat sources, and helps to detect fire signs and abnormal environmental changes in the early stage. The infrared radiation intensity at the current time point is collected in real time by infrared sensors (such as infrared beam sensors) and compared with the infrared radiation intensity at the previous time point. That is, |infrared radiation intensity at the current time point - infrared radiation intensity at the previous time point| / infrared radiation intensity at the previous time point. The result is the infrared blocking perturbation rate value. It should be noted that the infrared blocking perturbation rate value at the first time point is set to 0.

[0048] The specific steps for analyzing the environmental fire hazard assessment set of each area within the shopping mall's wireless network are as follows: Based on Z-score standardization, the combustible gas disturbance index and air ion concentration fluctuation index of each area within the shopping mall's wireless network are weighted (i.e., Z-score standardization is applied to the combustible gas disturbance index and air ion concentration fluctuation index of the corresponding area, and then weighted based on the results) to obtain the combustible factor anomaly index of the corresponding area; Based on Z-score standardization, the air temperature value, smoke concentration value, and infrared blocking disturbance rate value of each area within the shopping mall's wireless network are weighted (i.e., Z-score standardization is applied to the air temperature value, smoke concentration value, and infrared blocking disturbance rate value of the corresponding area, and then weighted based on the results) to obtain the combustible factor anomaly index of the corresponding area.

[0049] This implementation plan comprehensively evaluates multiple environmental monitoring data to assess potential fire-inducing environmental factors in various areas of the shopping mall. Secondly, by standardizing the environmental monitoring data using Z-scores, indicators with different dimensions and numerical scales can be compared and analyzed on the same basis, thereby improving the effectiveness of data fusion processing. Finally, through weighted processing, a fire hazard environmental disturbance index is generated, enabling the shopping mall to issue early warnings for potentially risky areas before a fire manifests as a visible flame or alarm signal. This enhances the proactiveness of fire hazard identification in the complex environment of the shopping mall and allows for timely fire prevention and control, effectively preventing the occurrence or escalation of fire accidents.

[0050] Specifically, such as Figure 2 As shown, the mall surveillance image angle sequence data specifically includes mall surveillance image data for each angle, including the pixel value and two-dimensional coordinates of each pixel. The specific steps for analyzing the visual safety hazard index of each area within the mall's wireless network are as follows: Input the mall surveillance image data of each area within the mall's wireless network into a pre-trained fire safety recognition model for comprehensive analysis to obtain the visual hazard assessment set for the corresponding area, including the visual thermal disturbance perception index and the fire-fighting component deterioration index; Based on the visual hazard assessment set for each area within the mall's wireless network, analyze the visual safety hazard index of the corresponding area.

[0051] The specific formula for calculating the visible security risk index of a certain area within a shopping mall's wireless network is as follows: Among them, KaQ is the visual safety hazard index of a certain area in the shopping mall wireless network, SrD is the visual thermal interference perception index of a certain area in the shopping mall wireless network, α1 is the thermal interference coefficient stored in the database, XgJ is the fire protection component deterioration index of a certain area in the shopping mall wireless network, α2 is the deterioration coefficient stored in the database, and α3 is the interaction coefficient stored in the database.

[0052] It should be explained that the specific steps for obtaining the thermal disturbance coefficient α1 stored in the database are as follows: Based on the multi-angle maximum response fusion layer, the fused optical scattering perturbation features, fused flame spectral texture frequency features, and fused smoke particle texture features of the corresponding regions are extracted, and the mean and standard deviation of the thermal disturbance features of the corresponding regions are extracted respectively. A ratio analysis is performed, such as |fused optical scattering perturbation features - mean thermal disturbance features| / standard deviation of thermal disturbance features. The ratio results are then weighted and averaged, and the result is used as the thermal disturbance coefficient α1.

[0053] The specific steps for obtaining the degradation coefficient α2 stored in the database are as follows: Based on the multi-angle matching fusion layer, extract the surface crack features, material aging features, and shading features of the fused fire protection facilities in the corresponding area, and extract the mean and standard deviation of the degradation features in the corresponding area respectively, and perform ratio analysis, such as |Surface crack features of fused fire protection facilities - mean of degradation features| / standard deviation of degradation features. The ratio results are then weighted and averaged, and the result is used as the degradation coefficient α2.

[0054] The specific steps for obtaining the interaction coefficient α3 stored in the database are as follows: Read the visual thermal disturbance perception index and fire component deterioration index of the area, and extract the visual sum value and visual interaction value (i.e., visual thermal disturbance perception index × fire component deterioration index) respectively. Then, the visual thermal disturbance perception index and fire component deterioration index are compared with the visual sum value respectively. The result of the ratio processing is weighted with the visual interaction value, and the result is used as the interaction coefficient α3.

[0055] The following is a specific implementation example for calculating the visible security hazard index of a certain area within a shopping mall's wireless network. The available data includes the visual thermal disturbance perception index and fire-fighting component deterioration index for five areas within the shopping mall's wireless network. (Specific details are as follows...) Figure 3-4 As shown:

[0056] Table 1. Example of visual hazard assessment set area sequence data within the shopping mall's wireless network.

[0057] Visual thermal disturbance perception index Fire protection component deterioration index Area 1 0.152 0.213 Area 2 0.284 0.317 Area 3 0.246 0.186 Area 4 0.172 0.481 Area 5 0.239 0.687

[0058] The thermal disturbance coefficient α1 stored in the database is approximately 0.412;

[0059] The degradation coefficient α2 stored in the database is approximately 0.389;

[0060] The interaction coefficient α3 stored in the database is approximately 0.543;

[0061] Substituting the data from Table 1 and the aforementioned coefficients into the specific formula for calculating the visible security hazard index of a certain area within the shopping mall's wireless network, we obtain:

[0062] The visual security risk index of the first area within the mall's wireless network is set to ln(1+0.152^0.412+0.213^0.389) / (1+0.543×exp(-0.152×0.213))≈0.457;

[0063] The visible security risk index of the second area within the mall's wireless network is set to ln(1+0.284^0.412+0.317^0.389) / (1+0.543×exp(-0.284×0.317))≈0.537;

[0064] The visible security risk index of the third area within the mall's wireless network is set to ln(1+0.246^0.412+0.186^0.389) / (1+0.543×exp(-0.246×0.186))≈0.468;

[0065] The visible security hazard index of the fourth area within the mall's wireless network is set to ln(1+0.172^0.412+0.481^0.389) / (1+0.543×exp(-0.172×0.481))≈0.534;

[0066] The visual security hazard index of the fifth area within the mall's wireless network is set to ln(1+0.239^0.412+0.687^0.389) / (1+0.543×exp(-0.239×0.687))≈0.604.

[0067] The fire safety identification model includes a feature extraction subnetwork, a feature fusion subnetwork, and an evaluation output subnetwork. The specific steps to obtain the visual hazard assessment set for each area within the designated shopping mall wireless network are as follows: In the feature extraction subnetwork of the fire safety identification model, the mall monitoring image data of each angle of each area within the designated shopping mall wireless network is received and processed to obtain the feature vector set of the corresponding angle. Specifically, each feature extraction layer in the feature extraction subnetwork extracts the corresponding features, such as optical scattering disturbance features, flame spectral texture frequency features, smoke particle texture features, fire protection facility surface crack features, fire protection facility material aging features, and fire protection facility occlusion features, and marks them as feature vector sets.

[0068] In the feature fusion subnetwork of the fire safety identification model, the feature vector set of each angle of each area within the shopping mall wireless network is classified and fused to obtain the joint feature vector set of the corresponding area. Specifically, the features of each angle are extracted by the multi-angle maximum response fusion layer and the multi-angle matching fusion layer, and then fused to generate fused optical scattering perturbation features, fused flame spectral texture frequency features, fused smoke particle texture features, fused fire facility surface crack features, fused fire facility material aging features, and fused fire facility occlusion features, and these features are marked as the fused feature vector set.

[0069] In the evaluation output sub-network of the fire safety identification model, the joint feature vector set of each area in the shopping mall wireless network is set for mapping processing to obtain the visual hazard assessment set of the corresponding area. Specifically, the corresponding fusion features are extracted based on the visual thermal disturbance perception layer and the fire component deterioration layer, and weighted processing is performed to output the visual thermal disturbance perception index and the fire component deterioration index.

[0070] The feature extraction subnetwork includes an optical scattering perturbation feature extraction layer, a flame spectral texture frequency feature extraction layer, a smoke particle texture feature extraction layer, a fire protection facility surface crack feature extraction layer, a fire protection facility material aging feature extraction layer, and a fire protection facility occlusion feature extraction layer.

[0071] Furthermore, the optical scattering perturbation feature extraction layer processes the pixel values ​​of each pixel in the shopping mall monitoring image at each angle of each region into grayscale values, applies the Sobel operator to calculate the gradient magnitude of each pixel, and divides the entire image into multiple non-overlapping square sub-blocks. Based on the two-dimensional Fast Fourier Transform (FFT) algorithm, the gradient magnitude of each pixel in each square sub-block is calculated to extract frequency domain information, i.e., the energy in each frequency region (coordinates). The total energy in the high-frequency region (the region accounting for more than 70% of the entire frequency range) of the corresponding square sub-block is counted, and the total frequency energy of the corresponding square sub-block (i.e., the sum of energy values ​​in all frequency regions) is also counted. A ratio is then calculated (e.g., the total energy in the high-frequency region of a certain square sub-block / the total frequency energy of the corresponding square sub-block), and a weighted average is performed based on the ratio calculation results to extract optical scattering perturbation features, which reflect the intensity of optical scattering caused by particulate matter such as smoke and dust in the air of the monitoring area.

[0072] The flame spectral texture frequency feature extraction layer processes the pixel values ​​of each pixel in the shopping mall surveillance image at each angle for each region, converting them into hue (H), saturation (S), and brightness (V) values ​​for the corresponding pixel. Based on the typical color temperature characteristics of flames (red-orange-yellow), a flame color threshold range is set, for example: hue (H) ∈ [0°, 50°], saturation (S) ≥ 0.4, brightness (V) ≥ 0.5. It then determines whether each pixel meets the above conditions, generating a binary mask image (flame candidate region mask). Pixels with a value of 1 in the mask represent flame candidate regions. Each pixel in the flame candidate region is extracted using a local binary model. The Low-Bandwidth (LBP) coding algorithm extracts texture features (LBP values) and generates an LBP-coded image of equal size. This image is then subjected to a two-dimensional discrete wavelet decomposition, resulting in a low-frequency approximation image and a high-frequency detail image (comprising three high-frequency sub-bands: horizontal, vertical, and diagonal high-frequency components). The low-frequency approximation image is then repeatedly subjected to multi-level wavelet decomposition to obtain the high-frequency components of each high-frequency sub-band at each corresponding angle. These components are then averaged to obtain the energy value of each layer, and a weighted average is applied to extract the flame spectral texture frequency features. Higher values ​​indicate a greater likelihood of typical flame high-frequency texture features in the current image.

[0073] The smoke particle texture feature extraction layer performs grayscale processing on the pixel values ​​of corresponding pixels, converting them into grayscale pixel values. Based on the gray-level co-occurrence matrix, it extracts contrast sub-features, energy sub-features, and homogeneity sub-features. Simultaneously, it uses the Sobel operator to calculate the gradient of the grayscale image to obtain the gradient magnitude of each pixel. Then, the image is divided into multiple identical sub-blocks, and the proportion of pixels in each sub-block with gradient magnitudes lower than a set threshold (e.g., 10) is counted and averaged to obtain the edge blurriness sub-feature. This sub-feature is then weighted with the contrast sub-feature, energy sub-feature, and homogeneity sub-feature to extract smoke particle texture features. The higher the value, the more obvious the suspected smoke texture structure in the image.

[0074] The surface crack feature extraction layer for fire protection facilities uses mall surveillance image data from the corresponding angle. Based on a target detection algorithm (such as YOLOv5), it extracts the region blocks (i.e., a set of pixels identified as fire protection facilities by the target detection algorithm, which are then marked as fire protection pixels) in the mall surveillance image where the fire protection facilities are located. Grayscale processing is then performed to generate grayscale pixel values ​​for each fire protection pixel. A Laplacian sharpening filter is applied for sharpening to obtain the sharpened grayscale pixel values ​​for each fire protection pixel. Finally, the Canny edge detection algorithm is used to extract the crack edge region (i.e., the set of all fire protection pixels identified as belonging to the crack edge, which are then marked as crack edges). The image processing algorithm first extracts the crack edge pixels and then uses morphological dilation and closing operations to expand the crack edge region to restore the complete crack region (including several crack edge pixels). This yields a binary edge map of the crack region. Next, a skeletonization algorithm is used to extract a set of crack edge pixels representing connected lines of single-pixel width within the crack region. This set is located at the center of the crack region. Finally, a contour search algorithm (such as findContours in OpenCV) is used to extract several sets of crack edge contours (i.e., sets of crack edge pixels).

[0075] Principal Component Analysis (PCA) is performed on the edge contour set of each crack region to obtain the principal axis direction (i.e., the first principal component direction and its corresponding direction vector). Using the principal axis direction as a reference, the maximum projection distance of the contour point set in this direction (i.e., the maximum minus the minimum value of the projection points along the principal axis direction) is calculated. This distance is the crack length. The number of crack fire protection pixels in each crack region is counted to obtain the area. Based on the direction vector of the first principal component direction of each crack region, the corresponding crack direction angle (i.e., the angle between the first principal component direction and the horizontal line) is analyzed. The mean crack length, mean area, and crack direction dispersion (i.e., standard deviation) are extracted and standardized. Based on the results of this standardization, a weighted average is applied to extract the surface crack features of the fire protection facility. A higher number indicates that the cracks are more severe or more densely distributed, affecting the reliability of the equipment in fire response.

[0076] The fire protection facility material aging feature extraction layer reads the area block in the mall's surveillance image where the fire protection facility is located, and performs color conversion on the pixel value of each fire protection pixel to obtain L (brightness), a (green-red), and b (blue-yellow) of the corresponding fire protection pixel. The mean value is then taken and compared with the preset normal material reference value (the corresponding reference mean value established from historical normal image samples) for difference processing (e.g., the square root of the sum of the squares of the differences between the mean values ​​of L, a, and b and the corresponding reference mean values) to obtain the color difference value. Based on the gray-level co-occurrence matrix (GLCM) algorithm, the image texture features of the area block are extracted, including equipment contrast, equipment entropy, and equipment homogeneity. Combined with the color difference value, the features are standardized and then weighted to extract the fire protection facility material aging features. The larger the value, the more serious the aging of the equipment material. This value is used to help identify safety hazards such as surface aging, paint peeling, corrosion, and rust.

[0077] The fire protection facility occlusion feature extraction layer reads the area blocks in the mall's surveillance image where the fire protection facilities are located. Based on the instance segmentation model, all non-fire protection facility foreground elements in the image are further segmented to extract an occlusion candidate region set (i.e., a set of pixels analyzed as foreground elements based on the instance segmentation model and marked as occluded pixels; foreground elements include, but are not limited to, people, goods, advertising boards, decorations, etc.). Pixel-level Boolean intersection is performed on each fire protection pixel in the area block and the occluded pixels in each occlusion candidate region to generate the pixel overlap area of ​​each occluded region on the area block. The total occlusion area is then summed to obtain the total occlusion area. The total fire protection pixel in the area block is counted and marked as the total equipment area area. A ratio is then applied: total occlusion area / total equipment area, to extract the fire protection facility occlusion features, which reflect the impact of the degree of occlusion of the mall's fire protection facilities on visual recognizability.

[0078] The feature fusion subnetwork includes a multi-angle maximum response fusion layer and a multi-angle matching fusion layer.

[0079] The multi-angle maximum response fusion layer uses the maximum response method to fuse features for optical scattering perturbation features, flame spectral texture frequency features, and smoke particle texture features at each angle. Specifically, it calculates the maximum value of optical scattering perturbation features, flame spectral texture frequency features, and smoke particle texture features at each angle and uses these values ​​as the fused optical scattering perturbation features, fused flame spectral texture frequency features, and fused smoke particle texture features.

[0080] The multi-angle matching and fusion layer involves matching and fusing the surface crack features, material aging features, and obstruction features of fire protection facilities from each angle. This includes statistically analyzing the mean and standard deviation of surface crack features and performing a ratio analysis: 1 - standard deviation of surface crack features / mean of surface crack features to obtain a matching value. This value is then weighted with the surface crack features from each angle: matching value × exp[-|surface crack features at a certain angle - mean of that feature| / standard deviation of that feature]. The results of this process are summed for each angle, and the summation is compared to obtain a weighting coefficient for each angle's surface crack features. This coefficient is then used for weighting, and the weighted results are mapped using a sigmoid function to obtain the fused surface crack features of the fire protection facilities.

[0081] The evaluation output sub-network includes a visual thermal disturbance perception layer and a fire protection component deterioration layer.

[0082] The visual thermal disturbance perception layer extracts the fused optical scattering perturbation features, fused flame spectral texture frequency features, and fused smoke particle texture features of each region, and performs weighted processing. Based on the weighted processing results, the sigmoid function is used for mapping, and the output is a visual thermal disturbance perception index in the range of 0-1, which is used to characterize the intensity of visually perceived thermal anomalies and the probability of fire signs.

[0083] The deterioration layer of fire protection components is obtained by extracting the surface crack features, material aging features, and occlusion features of the fire protection facilities in each area and weighting them. The weighted results are then mapped using the sigmoid function to output a fire protection component deterioration index ranging from 0 to 1. This index is used to characterize the overall deterioration status and potential failure risk of fire protection facility components within the monitoring area.

[0084] The pre-training steps for the fire safety identification model are as follows:

[0085] In the pre-training phase of the fire safety identification model, a training dataset covering multiple types of fire safety visual risk features is first constructed. Historical monitoring image data from typical public places such as shopping malls and office buildings are collected. The collected images cover various visual safety hazard states and environmental disturbance scenarios. Data categories include, but are not limited to: fire facility status image data (such as normal equipment, obstruction, displacement, damage, cracks, aging, etc.), visual thermal anomaly image data (such as flame color temperature area, smoke scattering area, abnormal image contrast, etc.), and real-time disturbance scenario data (such as crowding, abnormal stacking, environmental shadows, infrared obstruction, etc.). For different types of image data, data processing operations such as image enhancement, multi-angle combined sampling (simulating multiple camera perspectives) and visual feature labeling (including crack level, obstruction density, thermal disturbance intensity, etc.) are performed.

[0086] After completing the training data preparation, each sub-network in the fire safety identification model is independently pre-trained. Supervised learning is adopted, and image feature samples are input into the corresponding networks to output known visual risk labels. The loss function in the pre-training stage is either the cross-entropy loss function (for classification sub-tasks) or the mean squared error loss function (for continuous value output tasks). The training optimizer is the Adam optimizer to gradually optimize the network parameters and capture the sensitive patterns of various fire visual risk features.

[0087] After completing the independent pre-training of each sub-network, the weight parameters learned in each sub-network are loaded into the complete fire safety recognition model as initial weights, and the multi-level visual feature fusion optimization stage is entered. In this stage, multi-angle image inputs are uniformly synchronized to the fusion structure, the outputs of each feature layer are extracted and sent to the feature fusion sub-network. By constructing an angle consistency scoring function, the multi-view features are weighted and fused. Subsequently, the fused features are input into the evaluation output sub-network to predict: visual thermal disturbance perception index and fire component deterioration index respectively. Each output layer adopts an independent weight update mechanism and loss calculation process, and jointly minimizes the overall visual hazard assessment error.

[0088] Through the training and optimization of the above stages, the fire safety identification model can accurately identify various fire safety hazards in the face of different shopping mall areas, different camera angles and various visual interferences. It also has the comprehensive capabilities of multi-target feature recognition, fusion modeling and index prediction, thereby meeting the engineering requirements of real-time visual safety assessment tasks.

[0089] In this implementation plan, by comprehensively analyzing the mall's surveillance image data, safety hazards in each area can be accurately identified. First, using multi-angle surveillance image data avoids blind spots caused by single-angle perspectives, thereby improving the accuracy of monitoring and identification. Furthermore, the extraction of different features in the images allows for a more detailed assessment of potential safety hazards. Second, the combination of thermal disturbance perception index and fire protection component deterioration index not only measures fire risk but also reflects the status of fire protection facilities, thus clearly determining the severity of hazards and issuing warnings at different levels. Finally, the mall can achieve dynamic and real-time security monitoring, which not only improves the efficiency of security prevention but also reduces human intervention errors, thereby enhancing overall security.

[0090] Please see Figure 5 This invention provides a technical solution: a fire safety early warning system based on wireless networking, comprising: an electrical fire hazard analysis module, used to acquire in real time power monitoring data and environmental monitoring data of several areas within a set wireless network of a shopping mall, and analyze the electrical fire hazard induction index of the corresponding area; an environmental fire hazard analysis module, used to acquire environmental monitoring data of each area within the set wireless network of the shopping mall, and analyze the fire environmental disturbance index of the corresponding area; a visual safety hazard analysis module, used to acquire the angle sequence data of shopping mall monitoring images of each area within the set wireless network of the shopping mall, and combine it with a pre-trained fire safety identification model for comprehensive analysis to obtain the visual safety hazard index of the corresponding area; a comprehensive safety perception early warning module, used to analyze the comprehensive safety perception index of the corresponding area based on the electrical fire hazard induction index, fire environmental disturbance index, and visual safety hazard index of each area within the set wireless network of the shopping mall; and a fire safety early warning feedback module, used to provide fire safety early warning for the corresponding area based on the comprehensive safety perception index of each area within the set wireless network of the shopping mall.

[0091] Although preferred embodiments of the invention have been described, those skilled in the art, upon learning the basic inventive concept, can make other changes and modifications to these embodiments. Therefore, the appended claims are intended to be interpreted as including both the preferred embodiments and all changes and modifications falling within the scope of the invention.

[0092] Obviously, those skilled in the art can make various modifications and variations to this invention without departing from its spirit and scope. Therefore, if these modifications and variations fall within the scope of the claims of this invention and their equivalents, this invention also intends to include these modifications and variations.

Claims

1. A fire safety early warning method based on wireless networking, characterized in that, Includes the following steps: Real-time acquisition of power monitoring data and environmental monitoring data for several areas within the designated shopping mall wireless network, and analysis of the electrical fire hazard induction index and fire environmental disturbance index for the corresponding areas; The process involves acquiring angular sequence data of surveillance images from each area within a designated shopping mall wireless network. Specifically, this includes surveillance image data for each angle, containing the pixel value and two-dimensional coordinates of each pixel. This data is then comprehensively analyzed in conjunction with a pre-trained fire safety identification model. This model comprises a feature extraction subnetwork, a feature fusion subnetwork, and an evaluation output subnetwork to obtain the visual safety hazard index for the corresponding area. The specific steps are as follows: The mall's surveillance image data for each area within the designated wireless network is input into a pre-trained fire safety identification model for comprehensive analysis, resulting in a visual hazard assessment set for the corresponding area, including a visual thermal disturbance perception index and a fire-fighting component deterioration index, specifically: In the feature extraction subnetwork of the fire safety identification model, the mall monitoring image data of each area and each angle within the set mall wireless network is received and extracted to obtain the feature vector set of the corresponding angle. In the feature fusion subnetwork of the fire safety identification model, the feature vector set of each angle of each area within the shopping mall wireless network is classified and fused to obtain the joint feature vector set of the corresponding area. In the evaluation output subnetwork of the fire safety identification model, the joint feature vector set of each area in the shopping mall wireless network is set for mapping processing to obtain the visual hazard assessment set of the corresponding area. Based on the visual hazard assessment set of each area within the shopping mall's wireless network, the visual security hazard index of the corresponding area is analyzed. Based on the electrical fire hazard induction index, fire environment disturbance index, and visible safety hazard index of each area within the shopping mall's wireless network, the comprehensive safety perception index of the corresponding area is analyzed. The specific formula is as follows: Among them, ZaG, DqH, HxJ, and KaQ are the comprehensive safety perception index, electrical fire hazard induction index, fire environment disturbance index, and visible safety hazard index of a certain area within the shopping mall wireless network, respectively; β1, β2, β3, β4, and ω are the electrical coefficient, environmental coefficient, visibility coefficient, electrical environment coordination coefficient, and adjustment coefficient stored in the database, respectively. Based on the comprehensive security perception index of each area within the shopping mall's wireless network, fire safety warnings are issued for the corresponding areas.

2. The fire safety early warning method based on wireless networking according to claim 1, characterized in that, The power monitoring data includes power load thermal index, cable leakage micro-current value, contact resistance drift value, load connection surge intensity value, spike pulse interference intensity value, and arc disturbance value. The specific steps for analyzing and setting the electrical fire hazard induction index for each area within the shopping mall's wireless network are as follows: Based on the power monitoring data of each area within the shopping mall's wireless network, the electrical fire hazard assessment set of the corresponding area is analyzed, including the electrical overload thermal change index and the electrical transient anomaly index. Based on the electrical fire hazard assessment set of each area within the shopping mall's wireless network, the electrical fire hazard induction index of the corresponding area is analyzed.

3. The fire safety early warning method based on wireless networking according to claim 2, characterized in that, The specific steps for analyzing and setting up the electrical fire hazard assessment set for each area within the shopping mall's wireless network are as follows: Based on Z-score standardization, the power load thermal index, cable leakage microcurrent value, and contact resistance drift value of each area in the shopping mall wireless network are weighted to obtain the electrical overload thermal index of the corresponding area. Based on Z-score standardization, the load surge intensity value, spike pulse interference intensity value, and arc disturbance value of each area in the shopping mall wireless network are weighted to obtain the electrical transient anomaly index of the corresponding area.

4. The fire safety early warning method based on wireless networking according to claim 1, characterized in that, The environmental monitoring data includes combustible gas disturbance index, air temperature value, smoke concentration value, air ion concentration fluctuation index, and infrared blocking disturbance rate value. The specific steps for analyzing and setting the fire hazard environmental disturbance index for each area within the shopping mall's wireless network are as follows: Based on the environmental monitoring data of each area within the shopping mall's wireless network, the environmental fire risk assessment set of the corresponding area is analyzed, including the abnormal index of combustible factors and the abnormal index of thermal radiation. Based on the environmental fire risk assessment set of each area within the shopping mall's wireless network, the fire risk environmental disturbance index of the corresponding area is analyzed.

5. The fire safety early warning method based on wireless networking according to claim 4, characterized in that, The specific steps for analyzing and setting up the environmental fire hazard assessment set for each area within the shopping mall's wireless network are as follows: Based on Z-score standardization, the combustible gas disturbance index and air ion concentration fluctuation index of each area in the shopping mall wireless network are weighted to obtain the corresponding combustible factor abnormal index of the area. Based on Z-score standardization, the air temperature value, smoke concentration value, and infrared blocking disturbance rate value of each area in the shopping mall wireless network are weighted to obtain the corresponding area's combustible factor abnormality index.

6. The fire safety early warning method based on wireless networking according to claim 1, characterized in that, The specific formula for calculating the visible security risk index of a certain area within a shopping mall's wireless network is as follows: Among them, KaQ, ​​SrD, and XgJ are the visible security hazard index, visual thermal disturbance perception index, and fire protection component deterioration index of a certain area within the shopping mall wireless network, respectively, while α1, α2, and α3 are the thermal disturbance coefficient, deterioration coefficient, and interaction coefficient stored in the database, respectively.

7. A fire safety early warning system based on wireless networking, employing the fire safety early warning method based on wireless networking as described in any one of claims 1-6, characterized in that, include: The electrical fire hazard analysis module is used to acquire power monitoring data and environmental monitoring data of several areas within the set shopping mall wireless network in real time, and analyze the electrical fire hazard induction index of the corresponding areas. The environmental fire hazard analysis module is used to acquire environmental monitoring data for each area within the designated shopping mall wireless network and analyze the fire hazard environmental disturbance index of the corresponding area. The visual security hazard analysis module acquires the angle sequence data of the mall surveillance images of each area within the set mall wireless network, and performs comprehensive analysis in combination with the pre-trained fire safety identification model to obtain the visual security hazard index of the corresponding area. The comprehensive safety perception and early warning module is used to analyze the comprehensive safety perception index of each area based on the electrical fire hazard induction index, fire environment disturbance index, and visible safety hazard index of each area within the shopping mall's wireless network. The fire safety early warning feedback module is used to provide fire safety early warnings for each area within the shopping mall's wireless network based on the comprehensive safety perception index of that area.

Images