A fire safety management method and system based on a structured algorithm of AI cameras in a specific area

Abstract

The present invention relates to the field of image data processing technology, and provides a fire safety management method and system based on a structured algorithm of an AI camera in a specific area, including: collecting images of a specific area based on an AI camera to obtain image data; analyzing the image data based on the structured algorithm in the AI ​​camera to obtain fire safety features; obtaining current environmental information, inputting the environmental information into an offset model for calculation, and obtaining a corresponding offset value; offsetting a preset parameter of a preset fire safety prediction model based on the offset value to obtain a fire safety prediction model after offset; inputting the fire safety features into the fire safety prediction model after offset for prediction to obtain a fire safety prediction result, and generating a corresponding fire safety management plan based on the fire safety prediction result. In the present invention, the prediction result of fire safety is made more accurate.

Description

Technical Field

[0001] The present invention relates to the technical field of image data processing, and in particular to a fire safety management method and system based on a specific area AI camera structured algorithm. Background Art

[0002] At present, common fire safety management methods mainly include traditional manual inspections and installation of various simple sensor devices, such as smoke alarms, temperature sensors, etc. However, these methods have obvious disadvantages. Manual inspections consume a lot of manpower and are inefficient. It is difficult to achieve real-time and comprehensive monitoring of the area, and it is easy to miss and delay. Simple sensor devices can usually only detect a single physical parameter, and their reflection of complex fire safety conditions is very limited.

[0003] Although AI cameras are gradually being used in the field of fire safety, existing fire safety management solutions based on AI cameras are still insufficient. Most solutions simply use image recognition technology to detect flames or smoke, lacking in-depth analysis of fire safety conditions. At the same time, existing technologies largely ignore the important impact of environmental factors on fire safety. For example, environmental factors such as temperature, humidity, and wind direction can have a significant impact on the occurrence and development of fires and the effectiveness of firefighting measures. However, current fire safety management systems often fail to take these environmental factors into account, which leads to inaccurate predictions and judgments of fire safety, making it difficult to develop efficient and targeted fire safety management solutions. Summary of the invention

[0004] The main purpose of the present invention is to provide a fire safety management method and system based on a structured algorithm of an AI camera in a specific area, aiming to overcome the current defect that the influence of environmental factors is not taken into account and the prediction of fire safety is not accurate enough.

[0005] To achieve the above object, the present invention provides a fire safety management method based on a specific area AI camera structured algorithm, comprising the following steps:

[0006] Capturing images of a specific area based on an AI camera to obtain image data; analyzing the image data based on a structured algorithm in the AI ​​camera to obtain fire safety features;

[0007] Acquire current environmental information, input the environmental information into the offset model for calculation, and obtain a corresponding offset value;

[0008] offsetting the preset parameters of the preset fire safety prediction model based on the offset value to obtain the offset fire safety prediction model;

[0009] The fire safety feature is input into the offset fire safety prediction model for prediction to obtain a fire safety prediction result, and a corresponding fire safety management plan is generated based on the fire safety prediction result.

[0010] Furthermore, the fire safety characteristics include flame shape characteristics, smoke concentration characteristics, and heat source characteristics; the environmental information includes humidity information, wind information, and ambient temperature information.

[0011] Furthermore, the image data is analyzed based on the structured algorithm in the AI ​​camera to obtain fire safety features, including:

[0012] Performing color space conversion on the image data based on the structured algorithm in the AI ​​camera to extract color features of a specified color channel; the color features include flame color features and smoke color features;

[0013] Performing edge detection on the image data based on the structured algorithm in the AI ​​camera to obtain edge features; the edge features include flame edge features and smoke edge features;

[0014] Combining the flame color feature and the flame edge feature, a flame shape feature is obtained, and based on the smoke color feature and the smoke edge feature, a smoke concentration feature is obtained;

[0015] The image data is divided into different areas based on the structured algorithm in the AI ​​camera, and the temperature characteristics of each area are extracted as heat source characteristics.

[0016] Furthermore, the environmental information is input into the offset model for calculation to obtain the corresponding offset value, including:

[0017] The environmental temperature information, humidity information, and wind force information are respectively input into corresponding sub-offset models for calculation to obtain temperature offset values, humidity offset values, and wind force offset values; wherein each sub-offset model is trained based on the relationship between the corresponding environmental information and fire safety characteristics in historical data;

[0018] The temperature offset value, humidity offset value and wind force offset value are fused to obtain a final offset value.

[0019] Furthermore, based on the offset value, the preset parameters of the preset fire safety prediction model are offset to obtain the offset fire safety prediction model, including:

[0020] Based on the offset value, an offset calculation is performed on the connection weights between different neurons in the fire safety prediction model to obtain a fire safety prediction model after offset.

[0021] Further, after generating a corresponding fire safety management plan based on the fire safety prediction result, it includes:

[0022] Acquire multiple fire terminals associated with the specific area;

[0023] Negotiating an encrypted communication key with each of the fire-fighting terminals;

[0024] After the fire safety management plan is encrypted based on the encrypted communication key, it is distributed to each of the fire terminals; after each of the fire terminals receives the fire safety management plan, it reminds firefighters to perform fire management.

[0025] Further, the negotiating an encrypted communication key with each of the fire terminals includes:

[0026] Obtaining the fire management key of each fire terminal; sorting each fire terminal, and combining the fire management keys corresponding to each fire terminal in sequence according to the sorting to obtain a key string;

[0027] Obtain a binary tree structure template in a preset format, and add the characters in the key string to each node position of the binary tree structure template in sequence to obtain a key binary tree;

[0028] Obtaining the generation time of the fire safety management plan, and adding the numeric characters in the generation time to the key binary tree in sequence one by one to obtain a changed binary tree; wherein, starting from the root node of the key binary tree, each numeric character is added after the character at each node position to form a character combination;

[0029] Based on the changed binary tree, the encrypted communication key is generated.

[0030] Further, generating the encrypted communication key based on the changed binary tree includes:

[0031] Obtain the number corresponding to the number of the fire terminals as the first serial number; remove duplicate characters in the key string to obtain a deduplicated string, and obtain the number corresponding to the number of characters in the deduplicated string as the second serial number; obtain two preset serial numbers;

[0032] According to a preset order, a sequence number is set for each node position of the changed binary tree;

[0033] In the changed binary tree, nodes corresponding to the first sequence number, the second sequence number and two preset sequence numbers are obtained as target nodes, and nodes other than the target nodes in the changed binary tree are used as key nodes;

[0034] Connect each target node in sequence to obtain a plurality of straight lines; generate the encryption communication key according to the positional relationship between each key node and the plurality of straight lines.

[0035] The present invention also provides a fire safety management system based on a specific area AI camera structured algorithm, comprising:

[0036] An analysis unit, configured to collect images of a specific area based on an AI camera to obtain image data; and analyze the image data based on a structured algorithm in the AI ​​camera to obtain fire safety features;

[0037] A calculation unit, used to obtain current environmental information, input the environmental information into the offset model for calculation, and obtain a corresponding offset value;

[0038] An offset unit, configured to offset a preset parameter of a preset fire safety prediction model based on the offset value to obtain a fire safety prediction model after the offset;

[0039] A prediction unit is used to input the fire safety feature into the offset fire safety prediction model for prediction, obtain a fire safety prediction result, and generate a corresponding fire safety management plan based on the fire safety prediction result.

[0040] The present invention also provides a computer device, comprising a memory and a processor, wherein the memory stores a computer program, and the processor implements the steps of any one of the above methods when executing the computer program.

[0041] The present invention also provides a computer-readable storage medium on which a computer program is stored. When the computer program is executed by a processor, the steps of any of the above-mentioned methods are implemented.

[0042] The fire safety management method and system based on the structured algorithm of the AI ​​camera in a specific area provided by the present invention include: acquiring images of a specific area based on the AI ​​camera to obtain image data; analyzing the image data based on the structured algorithm in the AI ​​camera to obtain fire safety features; obtaining current environmental information, inputting the environmental information into the offset model for calculation, and obtaining a corresponding offset value; offsetting the preset parameters of the preset fire safety prediction model based on the offset value to obtain the offset fire safety prediction model; inputting the fire safety features into the offset fire safety prediction model for prediction to obtain fire safety prediction results, and generating a corresponding fire safety management plan based on the fire safety prediction results. In the present invention, by inputting environmental information into the offset model for calculation to obtain the corresponding offset value, and offsetting the preset parameters of the preset fire safety prediction model, the offset fire safety prediction model can more accurately predict and analyze the fire safety features, so that the fire safety prediction results are more accurate, thereby generating a reasonable fire safety management plan. BRIEF DESCRIPTION OF THE DRAWINGS

[0043] Figure 1 It is a schematic diagram of the steps of a fire safety management method based on a specific area AI camera structured algorithm in one embodiment of the present invention;

[0044] Figure 2 It is a structural block diagram of a fire safety management system based on a specific area AI camera structured algorithm in one embodiment of the present invention;

[0045] Figure 3 It is a schematic block diagram of the structure of a computer device according to an embodiment of the present invention.

[0046] The implementation, functional features and advantages of the present invention will be further described in conjunction with embodiments and with reference to the accompanying drawings. DETAILED DESCRIPTION

[0047] In order to make the purpose, technical solution and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are only used to explain the present invention and are not intended to limit the present invention.

[0048] Reference Figure 1 In one embodiment of the present invention, a fire safety management method based on a specific area AI camera structured algorithm is provided, comprising the following steps:

[0049] Step S1, collecting images of a specific area based on an AI camera to obtain image data; analyzing the image data based on a structured algorithm in the AI ​​camera to obtain fire safety features;

[0050] Step S2, obtaining current environmental information, inputting the environmental information into an offset model for calculation, and obtaining a corresponding offset value;

[0051] Step S3, offsetting the preset parameters of the preset fire safety prediction model based on the offset value to obtain the offset fire safety prediction model;

[0052] Step S4, inputting the fire safety feature into the offset fire safety prediction model for prediction, obtaining a fire safety prediction result, and generating a corresponding fire safety management plan based on the fire safety prediction result.

[0053] In this embodiment, as described in step S1 above, AI cameras are deployed in specific areas to achieve full coverage of the area. The above cameras have high-resolution and high-frame-rate image acquisition capabilities, and can capture various dynamic changes in the area in real time. For example, in a large shopping mall, AI cameras can be installed in various key locations, such as entrances, passages, stairwells, etc., to ensure no-dead-angle monitoring of the entire mall. The above specific areas are pre-set areas where fire problems are prone to occur.

[0054] The above AI cameras use image sensor technology to collect clear image data under different lighting conditions. Whether it is natural light during the day or artificial lighting at night, the image quality can be guaranteed, providing a reliable basis for subsequent analysis.

[0055] The above structured algorithm includes:

[0056] Color space conversion processing: First, the collected RGB image data is converted to other color spaces that are more suitable for analyzing fire safety features, such as HSV (hue, saturation, brightness) or YCbCr (brightness, blue chroma, red chroma). For example, in the HSV color space, the color of the flame is usually within a specific hue and saturation range. By extracting this feature information, potential fire sources can be identified more accurately. The converted color channels are analyzed to extract specific color distribution features related to fire. For example, a specific color threshold can be set. When the color value of a certain area in the image falls within this threshold range, it is considered to be a potential fire source or smoke area.

[0057] Edge detection processing: Use a variety of edge detection algorithms, such as Sobel operator, Canny edge detection, etc., to process the image and determine the outline of the object. The above algorithms can detect areas in the image where the pixel values ​​change dramatically, that is, the edge parts. Analyze the detected edge features to determine whether there is an edge shape that is consistent with the spread of flames or smoke. For example, flames usually form irregular edges during the spread process, while the edges of smoke are relatively blurred. By identifying these edge features, it is possible to quickly determine whether there is a fire safety problem.

[0058] Region segmentation processing: Use image segmentation algorithms, such as threshold-based segmentation, region growing, etc. to divide the image into different regions. For example, you can perform threshold segmentation based on the grayscale value or color characteristics of the image to divide the image into multiple different regions. Analyze each segmented region to determine whether it has abnormal temperature or texture characteristics. For example, you can use the temperature information obtained by an infrared camera to determine whether the temperature of a certain area is abnormally high; or by analyzing the texture characteristics of the image, determine whether there are texture changes related to fire, such as texture blurring caused by smoke.

[0059] As described in step S2 above: by installing various environmental sensors in a specific area, current environmental information, including temperature, humidity, wind direction, etc., is obtained. These sensors can monitor changes in environmental parameters in real time and transmit data to the fire safety management system. For example, in a factory workshop, a temperature sensor, a humidity sensor, and a wind direction sensor can be installed to measure the temperature, humidity, and wind direction in the workshop, respectively. These sensors can be connected to the system by wire or wireless means to ensure timely transmission of data.

[0060] The above-mentioned offset model is a trained mathematical model that can calculate the corresponding offset value based on the input environmental information. This model is trained based on a large amount of historical data, which includes fire safety characteristics and prediction results under different environmental conditions. Different sub-offset models can be used for calculation for different categories of environmental information. For example, for temperature information, a sub-offset model based on the relationship between temperature and fire safety characteristics can be used; for humidity information, another sub-offset model based on the relationship between humidity and fire safety characteristics can be used. Each sub-offset model is obtained by analyzing and learning the relationship between the category of environmental information and fire safety characteristics in historical data. For example, machine learning algorithms, such as linear regression, neural networks, etc., can be used to train historical data to obtain a model that can accurately predict offset values.

[0061] As described in step S3 above, the preset fire safety prediction model usually contains multiple parameters, which are determined during the model training process and are used to control the behavior and performance of the model. For example, the weight parameter determines the degree of influence of different input features on the output results; the bias parameter is used to adjust the output baseline of the model; the threshold parameter is used to determine the boundary of the classification, etc. Before making parameter offset adjustments, it is necessary to determine the initial values ​​of these preset parameters. The above initial values ​​can be obtained by training on historical data, or they can be set based on experience or prior knowledge.

[0062] According to the offset value calculated in step S2, the preset parameters of the preset fire safety prediction model are offset adjusted. The specific adjustment method can be selected according to different parameter types and model structures. For example, for weight parameters, the offset value can be added or multiplied to the original weight value to obtain the adjusted weight value. If the offset value is a positive number, it means that in the current environment, the influence of the input feature on the output result increases; if the offset value is a negative number, it means that the influence decreases. For the offset parameter, the offset value can be directly added to the original offset value to adjust the output baseline of the model. For the threshold parameter, the size of the threshold can be appropriately adjusted according to the size and direction of the offset value to change the classification boundary of the model.

[0063] As described in the above step S4: the fire safety features obtained in step S1 are input into the fire safety prediction model after the offset. The above fire safety features are obtained by analyzing the image data, including flame shape features, smoke concentration features, and heat source features. The input features can be a single feature vector or a combination of multiple feature vectors. For example, the flame shape feature, smoke concentration feature, and heat source feature can be represented as a vector respectively, and then these vectors are combined into a higher-dimensional feature vector and input into the model for prediction.

[0064] The fire safety prediction model after the offset performs prediction calculations based on the input fire safety features and the adjusted parameters to obtain the fire safety prediction results. This result can be a numerical value indicating the degree of fire safety risk in the current specific area; it can also be a classification result, such as "safe", "low risk", "high risk", etc. The above prediction model can use a variety of algorithms, such as logistic regression, decision tree, support vector machine, neural network, etc. Different algorithms have different characteristics and applicable scenarios, and can be selected according to actual conditions.

[0065] Generate a corresponding fire safety management plan based on the prediction results. This plan can be a specific action plan or a set of suggestions and guidance. If the prediction results show that there is a high risk of fire, the plan may include specific measures such as immediately sounding an alarm, notifying the fire department and relevant responsible persons, starting firefighting equipment, and guiding personnel to evacuate. For example, an alarm signal can be issued through an audible and visual alarm, and the alarm information can be sent to the mobile phone or computer of the relevant personnel at the same time; fire-fighting facilities such as fire-fighting equipment and smoke exhaust equipment can be automatically started; evacuation routes and safe exit locations can be displayed on the display screen to guide personnel to evacuate quickly. If the prediction results show that the risk is low, some preventive measures can be taken, such as strengthening patrols and checking whether the fire-fighting equipment is operating normally.

[0066] In summary, the fire safety management method based on the structured algorithm of AI cameras in specific areas realizes comprehensive and accurate management of fire safety in specific areas through the collection and analysis of image data, the processing of environmental information, the adjustment of model parameters and the generation of prediction results, providing strong technical support for protecting people’s lives and property.

[0067] In one embodiment, the fire safety characteristics include flame shape characteristics, smoke concentration characteristics, and heat source characteristics; the environmental information includes humidity information, wind information, and ambient temperature information.

[0068] In one embodiment, the image data is analyzed based on the structured algorithm in the AI ​​camera to obtain fire safety features, including:

[0069] Performing color space conversion on the image data based on the structured algorithm in the AI ​​camera to extract color features of a specified color channel; the color features include flame color features and smoke color features;

[0070] Performing edge detection on the image data based on the structured algorithm in the AI ​​camera to obtain edge features; the edge features include flame edge features and smoke edge features;

[0071] Combining the flame color feature and the flame edge feature, a flame shape feature is obtained, and based on the smoke color feature and the smoke edge feature, a smoke concentration feature is obtained;

[0072] The image data is divided into different areas based on the structured algorithm in the AI ​​camera, and the temperature characteristics of each area are extracted as heat source characteristics.

[0073] In this embodiment, the original image data is usually represented in the common RGB (red, green, blue) color space, but this color space is not directly effective for identifying the color characteristics of flames and smoke. By converting the image data into a color space, the color characteristics of a specific color channel related to fire safety can be more accurately extracted. For example, by converting to the HSV (hue, saturation, value) color space, the hue can better reflect the type of color, and the saturation and value can help distinguish the vividness and brightness of the color. Flames and smoke usually have unique expressions within a specific HSV value range.

[0074] Based on the converted color space, a specific color threshold range is set to identify flame color features. Flames usually have obvious characteristics under a certain combination of hue, saturation, and brightness. For example, the hue of a flame is concentrated in a specific angle range, the saturation is relatively high, and the brightness also has a certain range. By analyzing the HSV value of each pixel in the image, it is determined whether it falls within the threshold range of the flame color feature. If so, the pixel is considered to possibly belong to the flame area.

[0075] Smoke also has its own unique performance in color space. Usually, smoke makes the color of the image more blurred and dim, with specific color distribution characteristics. Similarly, in the converted color space, by analyzing the color value distribution of pixels, the range of color characteristics that smoke may appear is determined. For example, smoke may reduce the saturation of the image and change the brightness. By setting the corresponding threshold, the color characteristics of smoke can be extracted.

[0076] The above-mentioned edge detection is an important step in image processing, which can help determine the outline and boundary of an object. In fire safety feature extraction, edge detection can be used to identify the edge features of flames and smoke, which are crucial for determining whether there is a fire hazard. The image is processed using edge detection algorithms, such as the Sobel operator, Canny edge detection, etc. Flames usually form irregular edges during the diffusion process, and these edges have specific shape and intensity characteristics. By analyzing the edge detection results, determine which edges may belong to the flame area. For example, the edge of a flame usually has a high intensity and an irregular shape, which is clearly different from the edges of the surrounding environment. Specific edge intensity and shape thresholds can be set to filter out flame edge features.

[0077] The edge of smoke is relatively fuzzy, not as obvious as the edge of flame. However, smoke will also make the edge of the image fuzzy and unclear during the diffusion process. The edge change can be detected by edge detection algorithm. For example, the edge of smoke may reduce the gradient value of the image and the continuity of the edge will also be affected. By analyzing these features, the smoke edge features can be extracted.

[0078] By combining the color features and edge features of the flame, the shape features of the flame can be determined more accurately. The color features of the flame can help determine the possible flame area, while the edge features can further refine the outline of the flame. By matching pixels that meet the flame color features with the flame edge features, a more accurate flame shape can be obtained. For example, if a pixel meets the threshold requirements of the flame color feature and is located in the area determined by the flame edge feature, then the pixel is considered to be part of the flame. By analyzing and combining multiple such pixels, the shape features of the flame can be obtained.

[0079] The color features and edge features of smoke can also be combined to determine the density features of smoke. The color features of smoke can provide preliminary information about the presence of smoke, while the edge features can reflect the degree of diffusion of smoke. When the smoke density is high, the color of the image will become more blurred and dim, and the edges will be less clear. By analyzing the degree of change of color features and edge features, the density of smoke can be estimated. For example, different density levels can be set to determine which density level the smoke belongs to based on the combination of color and edge features.

[0080] In this embodiment, the image data is divided into different regions using an image segmentation algorithm. Region segmentation can be performed based on the grayscale value, color features, texture features, etc. of the image. For example, a threshold-based segmentation method can be used to divide the image into multiple different regions, each region having similar color or texture features. Alternatively, a region growing algorithm can be used to gradually merge adjacent pixels into the same region starting from a seed point.

[0081] In each divided area, the temperature feature is extracted as the heat source feature. This can be estimated by combining the temperature information obtained by the infrared camera or using the color features of the image. For example, if the color of a certain area in the image is brighter or has a specific color distribution, it may mean that the temperature in that area is higher. By analyzing the temperature features of each area, potential heat source areas can be identified. These heat source areas may be potential fire hazards and require further attention and analysis.

[0082] In one embodiment, the environmental information is input into the offset model for calculation to obtain the corresponding offset value, including:

[0083] The environmental temperature information, humidity information, and wind force information are respectively input into corresponding sub-offset models for calculation to obtain temperature offset values, humidity offset values, and wind force offset values; wherein each sub-offset model is trained based on the relationship between the corresponding environmental information and fire safety characteristics in historical data;

[0084] The temperature offset value, humidity offset value and wind force offset value are fused to obtain a final offset value.

[0085] In this embodiment, first, the acquired environmental information is divided into environmental temperature information, humidity information and wind information. These three environmental factors play an important role in fire safety, and different factors will have different effects on the occurrence and development of fire and the effectiveness of fire fighting measures. For example, an increase in environmental temperature may increase the risk of fire, a change in humidity may affect the spread of fire, and wind will affect the diffusion direction of smoke and the spread speed of fire.

[0086] For different types of environmental information, the corresponding sub-offset model is used for calculation. The above sub-offset model is trained based on the relationship between the corresponding environmental information and fire safety characteristics in historical data. For the temperature sub-offset model, it learns the influence of temperature on fire safety by analyzing the association between different temperature values ​​and fire safety characteristics (such as flame color characteristics, smoke concentration characteristics, etc.) in a large amount of historical data. When the current ambient temperature information is input, the model can calculate the temperature offset value according to the learned rules. This value reflects the degree of influence of the current temperature on the fire safety prediction model. Similarly, the humidity sub-offset model and the wind sub-offset model are also trained based on the relationship between humidity information and wind information in historical data and fire safety characteristics, and can calculate humidity offset values ​​and wind offset values.

[0087] Since the impact of different environmental factors on fire safety is interrelated, the offset value of a single environmental factor may not fully reflect the comprehensive impact of the environment on fire safety. Therefore, it is necessary to fuse the temperature offset value, humidity offset value, and wind offset value to obtain a comprehensive final offset value. For example, in an environment with high temperature, low humidity, and strong wind, the risk of fire will increase significantly. At this time, it is necessary to comprehensively consider the impact of these three factors to adjust the parameters of the fire safety prediction model.

[0088] In this embodiment, a variety of fusion methods can be used, such as weighted average method, principal component analysis method, etc. Taking the weighted average method as an example, different weights can be assigned to the temperature offset value, humidity offset value and wind offset value according to the importance of different environmental factors in the historical data to fire safety. Then, each offset value is multiplied by the corresponding weight and added together to obtain the final offset value. The above final offset value can more accurately reflect the comprehensive impact of the current environment on fire safety, and provide a more reliable basis for the subsequent parameter adjustment of the fire safety prediction model.

[0089] In one embodiment, the preset parameters of the preset fire safety prediction model are offset based on the offset value to obtain the offset fire safety prediction model, including:

[0090] Based on the offset value, an offset calculation is performed on the connection weights between different neurons in the fire safety prediction model to obtain a fire safety prediction model after offset.

[0091] In this embodiment, the above-mentioned preset fire safety prediction model is generally a model based on a neural network or other machine learning algorithm. In a neural network, the connection weights between different neurons determine the degree of influence of input features on output results. For example, a simple neural network consists of an input layer, a hidden layer, and an output layer. The input layer receives data such as fire safety features, passes the data to the hidden layer for processing through the connection weights between different neurons, and finally generates a fire safety prediction result in the output layer.

[0092] The offset value is calculated by inputting environmental information into the offset model, which reflects the degree of influence of the current environment on the fire safety prediction model. This offset value can be used to adjust the parameters of the model to make it more suitable for specific environmental conditions. For example, if the ambient temperature is high, the humidity is low, and the wind is strong, the offset value can indicate that the model needs to pay more attention to fire safety characteristics related to high temperature, dryness, and strong wind, such as faster flame spread, change in smoke spread direction, etc.

[0093] Based on the offset value, the connection weights between different neurons in the fire safety prediction model are offset calculated. The specific calculation method can be determined according to the type of model and the specific algorithm. For example, in a neural network, an optimization algorithm such as the gradient descent method can be used to update the connection weights. First, the product or sum of the current connection weight and the offset value is calculated to obtain the adjusted connection weight. Then, the output results of the model are recalculated based on the new connection weight to evaluate the performance of the adjusted model. If the performance of the adjusted model is improved, that is, the prediction results are more accurate, then the new connection weights are retained; otherwise, the offset value can continue to be adjusted or other optimization methods can be used to further improve the model.

[0094] In this embodiment, by dynamically adjusting the connection weights according to environmental information, the fire safety prediction model can better adapt to different environmental conditions, thereby improving the generalization ability and accuracy of the model.

[0095] In one embodiment, after generating a corresponding fire safety management plan based on the fire safety prediction result, the method includes:

[0096] Acquire multiple fire terminals associated with the specific area;

[0097] Negotiating an encrypted communication key with each of the fire-fighting terminals;

[0098] After the fire safety management plan is encrypted based on the encrypted communication key, it is distributed to each of the fire terminals; after each of the fire terminals receives the fire safety management plan, it reminds firefighters to perform fire management.

[0099] In this embodiment, the specific area can be a building, a factory area, a commercial complex, etc. The fire terminals associated with the specific area usually include various fire equipment, monitoring equipment, and communication equipment related to fire management. For example, in a large shopping mall, the fire terminals include fire alarms, control terminals of automatic sprinkler systems, fire broadcast systems, fire monitoring cameras, etc. By determining these fire terminals associated with the specific area, it can be ensured that the fire safety management plan can be accurately communicated to the relevant fire equipment and personnel.

[0100] In this embodiment, the fire terminal information associated with a specific area can be obtained through network connection, wireless communication, etc. For example, using the Internet of Things technology, each fire terminal is connected to a central management system to obtain the status and information of the terminal in real time. The list of fire terminals and related information can also be obtained through a pre-established database or configuration file. This information may include the type, location, communication address, etc. of the terminal for subsequent communication and management.

[0101] In fire safety management, ensuring the security of communication is crucial. Since the fire safety management plan may contain sensitive information, such as the location of the fire, the degree of danger, etc., the communication needs to be encrypted to prevent the information from being stolen or tampered with. Negotiating encrypted communication keys is an effective way to ensure communication security. By negotiating keys with each fire terminal, a secure communication channel can be established to ensure that the fire safety management plan can be safely transmitted to each terminal.

[0102] Key negotiation algorithms, such as the Diffie-Hellman key exchange algorithm, can be used to negotiate encrypted communication keys with each fire terminal. The above algorithm allows both parties to negotiate a shared key over an insecure communication channel without knowing the other party's key in advance. The specific process is as follows: First, the sender and receiver generate a pair of public and private keys respectively. Then, the sender sends its public key to the receiver, and the receiver also sends its public key to the sender. Then, both parties use the other party's public key and their own private key to calculate a shared key. This key can be used to encrypt and decrypt communications.

[0103] The fire safety management plan is encrypted using the negotiated encryption communication key. Encryption can use a symmetric encryption algorithm or an asymmetric encryption algorithm, depending on actual needs and security requirements. For example, the AES (Advanced Encryption Standard) symmetric encryption algorithm can be used to input the fire safety management plan as plain text, and the encryption communication key can be used to encrypt it to obtain the ciphertext. In this way, even if it is stolen during the communication process, the attacker cannot easily decipher the content of the plan.

[0104] The encrypted fire safety management plan is distributed to each fire terminal through network connection, wireless communication, etc. The distribution can be carried out by broadcast, multicast or unicast, depending on the number and distribution of fire terminals. For example, if there are multiple fire terminals distributed in different locations, the encrypted plan can be sent to multiple terminals at the same time by multicast. If only one specific fire terminal needs to receive the plan, it can be sent by unicast.

[0105] After receiving the encrypted fire safety management plan, each fire terminal uses the negotiated encrypted communication key to decrypt the plan and obtain the plain text fire safety management plan. According to the decrypted fire safety management plan, the fire terminal takes corresponding measures to remind firefighters to carry out fire management. This includes sounding an alarm, displaying warning information, sending text messages or push notifications. For example, a fire alarm can sound a loud alarm to alert nearby people to the fire; a fire surveillance camera can display warning information on the display screen to indicate the location and degree of danger of the fire; and a fire broadcast system can play voice notifications to guide people to evacuate and take firefighting measures.

[0106] In one embodiment, the negotiating an encrypted communication key with each of the fire terminals includes:

[0107] Obtaining the fire management key of each fire terminal; sorting each fire terminal, and combining the fire management keys corresponding to each fire terminal in sequence according to the sorting to obtain a key string;

[0108] Obtain a binary tree structure template in a preset format, and add the characters in the key string to each node position of the binary tree structure template in sequence to obtain a key binary tree;

[0109] Obtaining the generation time of the fire safety management plan, and adding the numeric characters in the generation time to the key binary tree in sequence one by one to obtain a changed binary tree; wherein, starting from the root node of the key binary tree, each numeric character is added after the character at each node position to form a character combination;

[0110] Based on the changed binary tree, the encrypted communication key is generated.

[0111] In this embodiment, each fire terminal generally has a unique fire management key to ensure the communication security between the terminal and the central management system or other terminals. This key can be pre-assigned or generated when the device is installed or initialized. The fire management key can be used to encrypt and decrypt communication data to ensure that only authorized devices can access and process information related to fire safety management.

[0112] The fire management key of each fire terminal can be obtained in a variety of ways. For example, when the equipment is installed, the key can be stored in the internal memory of the terminal, and the key information can be transmitted to the central management system through a secure communication channel. Alternatively, the key can be uniformly generated and distributed by a key management server to ensure the security and uniqueness of the key. The key can also be transmitted in an encrypted manner to prevent the key from being stolen or tampered with during transmission. For example, using public key encryption technology, the key is encrypted with the public key of the recipient, and only the recipient with the corresponding private key can decrypt the key.

[0113] Sorting the various fire terminals can ensure that the order of the key combination is consistent, thereby ensuring the stability and reliability of the generated encrypted communication key. Sorting can be performed based on factors such as the device number, installation location, and priority of the fire terminal. For example, the fire terminals can be sorted in order from small to large device numbers, or sorted according to the location of the fire terminals in a specific area, so that each terminal can be quickly located and managed when needed.

[0114] According to the sorted order, the fire management keys corresponding to each fire terminal are combined in sequence to obtain a key string. This key string contains the key information of all fire terminals and will be used in the subsequent key generation process. For example, if there are three fire terminals, their fire management keys are "key1", "key2" and "key3" respectively, and they are combined in the order obtained after sorting, the key string obtained is "key1key2key3".

[0115] The binary tree structure template with a preset format provides a structured way to organize and process key information. The binary tree structure has efficient storage and retrieval performance, and can easily perform key insertion, deletion and search operations. The above binary tree structure template is designed in advance according to, for example, the depth of the binary tree, the number and distribution of nodes, etc. can be determined.

[0116] The characters in the key string are added to each node position of the binary tree structure template one by one in order to obtain a key binary tree. The specific adding method can be performed according to the traversal order of the binary tree, such as pre-order traversal, in-order traversal or post-order traversal. For example, for a simple binary tree structure template, the first character in the key string is first added to the root node position, then the second character is added to the left child node position, the third character is added to the right child node position, and so on, until all characters are added to the nodes of the binary tree.

[0117] The generation time of the fire safety management plan can be used as a dynamic factor to increase the randomness and security of the encrypted communication key. The numeric characters in the generation time can be combined with the characters in the key binary tree to form a more complex key structure. The above generation time can be accurate to the second level.

[0118] Add the numeric characters in the generation time to the key binary tree one by one in sequence to obtain a changed binary tree. The specific adding method is to start from the root node of the key binary tree, and add each numeric character after the character at each node position to form a character combination. For example, if the root node character of the key binary tree is "K" and the generation time is "20240627103055", then add the numeric character "2" after the root node character "K" to form a new character combination "K2". Then, in the same way, add the other numeric characters in the generation time to the corresponding node positions in sequence to obtain a changed binary tree.

[0119] Based on the change binary tree, multiple methods can be used to generate encryption communication keys. For example, the change binary tree can be hashed and the hash value can be used as the encryption communication key. Alternatively, the change binary tree can be subjected to specific encoding and conversion operations to obtain a key of fixed length. The generated encryption communication key should have sufficient length and randomness to ensure the security of communication. At the same time, the key generation process should be reversible so that the encrypted data can be decrypted when needed. The generated encryption communication key will be used for secure communication with each fire terminal. When sending the fire safety management plan, the central management system uses the key to encrypt the plan, and then sends the encrypted plan to each fire terminal. After receiving the encrypted plan, the fire terminal uses the same key to decrypt it and obtains the plaintext fire safety management plan.

[0120] In one embodiment, generating the encrypted communication key based on the changed binary tree includes:

[0121] Obtain the number corresponding to the number of the fire terminals as the first serial number; remove duplicate characters in the key string to obtain a deduplicated string, and obtain the number corresponding to the number of characters in the deduplicated string as the second serial number; obtain two preset serial numbers;

[0122] According to a preset order, a sequence number is set for each node position of the changed binary tree;

[0123] In the changed binary tree, nodes corresponding to the first sequence number, the second sequence number and two preset sequence numbers are obtained as target nodes, and nodes other than the target nodes in the changed binary tree are used as key nodes;

[0124] Connect each target node in sequence to obtain a plurality of straight lines; generate the encryption communication key according to the positional relationship between each key node and the plurality of straight lines.

[0125] In this embodiment, the number corresponding to the number of the fire terminals is used as the first serial number. This number reflects the total number of fire terminals involved in fire safety management in a specific area. For example, if there are 10 fire terminals, then the first serial number is 10.

[0126] Perform a deduplication operation on the characters in the key string to obtain a deduplication string. Then, obtain the number corresponding to the number of characters in the deduplication string as the second serial number. The deduplication operation can remove duplicate characters in the key string, making the generated serial number more concise and representative. For example, if the deduplication string has 5 different characters, then the second serial number is the number 5.

[0127] The two preset serial numbers may be numbers pre-set according to specific rules or requirements. These serial numbers may be related to factors such as the security level of the system, parameters of the encryption algorithm, etc. For example, the two preset serial numbers may be set to 15 and 20, which are used to participate in the calculation together with the first serial number and the second serial number in the subsequent key generation process.

[0128] The above preset order can be determined according to the structural characteristics of the binary tree and the requirements of key generation. For example, the order of pre-order traversal, in-order traversal or post-order traversal can be used to set the serial number for changing the position of each node of the binary tree. Pre-order traversal is to visit the root node first, then traverse the left subtree and the right subtree; in-order traversal is to traverse the left subtree first, then visit the root node, and finally traverse the right subtree; post-order traversal is to traverse the left subtree and the right subtree first, and finally visit the root node.

[0129] According to the preset order, set the serial number for each node position of the binary tree. Starting from the root node, assign a unique serial number to each node in turn. For example, if the order of pre-order traversal is adopted, the root node is first assigned serial number 1, then the left child node is assigned serial number 2, then the right child node is assigned serial number 3, and so on. In this way, each node has a clear serial number, which is convenient for positioning and operation in the subsequent key generation process.

[0130] In the change binary tree, the nodes corresponding to the first sequence number, the second sequence number, and the two preset sequence numbers are obtained as target nodes. These nodes will play a key role in the key generation process. For example, if the first sequence number is 10, the second sequence number is 5, and the two preset sequence numbers are 15 and 20, then find the nodes with sequence numbers 10, 5, 15, and 20 in the change binary tree and use them as target nodes. The nodes in the change binary tree except the target node are used as key nodes. These nodes will participate in the generation of encrypted communication keys together with the target node. The number of key nodes is usually more than the target nodes, and they contain more key information. For example, if the change binary tree has 30 nodes, 4 of which are target nodes, then the remaining 26 nodes are key nodes.

[0131] Connect each target node in sequence to obtain multiple straight lines. The order of connection can be determined according to the size of the target node serial number, or according to specific rules or requirements. For example, if the serial numbers of the target nodes are 5, 10, 15 and 20, they can be connected in order from small to large to obtain multiple straight lines. Generate an encrypted communication key based on the positional relationship between each key node and multiple straight lines. The specific generation method can be calculated and encoded based on factors such as the relative position and distance between the key node and the straight line. For example, the distance, angle and other information from the key node to the straight line can be encoded to obtain an encrypted communication key. Alternatively, the key can be generated by encoding based on the positional relationship such as whether the key node is above, below, left or right of the straight line. The encrypted communication key generated in this way has high randomness and security, and can effectively ensure the confidentiality of the fire safety management plan during transmission.

[0132] Specifically, in one embodiment, generating the encrypted communication key according to the positional relationship between each of the key nodes and the plurality of straight lines includes:

[0133] The shortest distance between each of the key nodes and each straight line is obtained, and it is determined whether the shortest distance is less than a preset distance. If so, the corresponding key node is used as a target key node; the characters on each target key node are further combined to obtain a key character combination, and the key character combination is screened according to preset rules to obtain the encrypted communication key.

[0134] In this embodiment, for each key node, the distance between it and each straight line is calculated. A geometric method can be used to calculate the shortest distance between a node and a straight line. For example, the shortest distance can be determined by calculating the vertical distance from the node to the straight line. A preset distance value is set, and this value can be adjusted according to the needs of the system. If the shortest distance between a key node and a straight line is less than the preset distance, then it is considered that the positional relationship between the node and the straight line is relatively close, which may have an important impact on the generation of encrypted communication keys. By comparing the shortest distance and the preset distance, key nodes that are relatively close to the positional relationship of the straight line can be screened out, and these nodes will be further processed to generate encrypted communication keys.

[0135] If the shortest distance between a key node and each straight line is less than the preset distance, the key node is used as the target key node. In this way, a group of nodes that are closely related to the straight line position can be screened out, and the characters on these nodes may contain important key information. The characters on each target key node are further combined to obtain a key character combination. This combination can be a simple string concatenation or a character combination operation performed according to specific rules. For example, the characters on the target key node can be concatenated according to their position order in the binary tree, or the characters can be combined using a specific encoding method to increase the complexity and security of the key.

[0136] The above-mentioned preset rules can be determined according to the requirements of the encryption algorithm, the security level of the system and the actual application scenario. These rules can include a specific order of characters, a specific character combination pattern or a specific transformation operation on characters. For example, the rules can be set to retain only characters in a specific position, remove duplicate characters, convert characters according to a specific encoding method, etc.

[0137] The key character combination is screened according to the preset rules. This process can remove unnecessary characters and enhance the randomness and security of the key. The screened character combination is the final generated encryption communication key. This key will be used to encrypt the fire safety management plan to ensure the confidentiality and integrity of the plan during transmission.

[0138] In summary, the above technical solution calculates the positional relationship between the key node and the straight line, screens out the target key node, combines characters and screens them, and finally generates an encrypted communication key, thereby providing effective protection for the secure transmission of the fire safety management solution.

[0139] In one embodiment, based on the changed binary tree, the step of generating the encrypted communication key specifically includes:

[0140] Starting from the root node of the modified binary tree, a leaf node is randomly selected as a starting point, and the nodes of the binary tree are traversed in a preset direction based on the starting point. During the traversal process, the positions of the nodes passed and the character combinations at the positions are recorded;

[0141] When the traversal returns to the starting point, multiple lines are generated according to the recorded node position information and character combinations; for every two adjacent node positions, they are connected to form a line segment, and the corresponding character combination is marked on the line segment;

[0142] Analyze and process all generated line segments, extract feature information of the length, direction and character combination of the line segments, encode the feature information and obtain a code value sequence;

[0143] According to the preset key generation rules, the coded value sequence is processed to generate the encrypted communication key.

[0144] In this embodiment, first, starting from changing the root node of the binary tree, a leaf node is randomly selected as the starting point. The purpose of this is to increase the randomness of the generation of the encryption communication key so that the key generated each time has a higher security. Since the leaf node is the terminal node of the binary tree, selecting the leaf node as the starting point can ensure that the traversal process can cover different parts of the entire binary tree.

[0145] Taking the selected starting point as the reference, the nodes of the binary tree are traversed in a preset direction. The preset direction can be clockwise, counterclockwise or other specific traversal order. During the traversal process, the node positions passed through and the character combinations at the positions are recorded. The node position can be represented by the node number, coordinates or other methods in the binary tree, and the character combination is the character information stored on the node. When the traversal returns to the starting point, according to the recorded node position information, for every two adjacent node positions, they are connected to form a line segment. In this way, multiple lines can be obtained, which will be used in the subsequent key generation process. The process of connecting adjacent nodes can use the method of geometric drawing, such as connecting two points with a straight line on a two-dimensional plane. The corresponding character combination is marked on the line segment formed by connecting adjacent nodes. These character combinations are the character information on the nodes recorded during the traversal process. Marking character combinations can make the line segment have more information and provide a basis for subsequent analysis and processing.

[0146] All generated line segments are analyzed and processed. The content of the analysis may include the length and direction of the line segment and the characteristics of the character combination marked on the line segment. The length of the line segment can be obtained by calculating the distance between two nodes, and the direction can be expressed as an angle or vector. The characteristics of the character combination may include the number, type, and number of repetitions of the characters.

[0147] Extract the feature information of the length, direction and character combination of the line segment. The above feature information will be used to generate the encryption communication key, which can reflect the structure of the binary tree and the distribution of characters, and increase the complexity and security of the key. For example, the length and direction of the line segment can be converted into digital representation, and the features of the character combination can be encoded to obtain a series of feature values.

[0148] The extracted feature information is encoded to obtain a sequence of encoded values. The encoding method can be selected according to specific needs and encryption algorithms, which can be simple numerical encoding, hash encoding, or other more complex encoding methods. The purpose of encoding is to convert the feature information into a form that is easy to process and store, while increasing the randomness and security of the key.

[0149] According to the preset key generation rules, the coded value sequence is processed to generate the encrypted communication key. The key generation rule can be based on a specific mathematical algorithm, logical operation or other encryption technology. For example, the coded value sequence can be hashed, XORed or key expansion algorithm can be used to obtain the final encrypted communication key. This key will be used to encrypt the fire safety management plan to ensure the security of the plan during transmission.

[0150] Reference Figure 2 In another embodiment of the present invention, a fire safety management system based on a specific area AI camera structured algorithm is provided, including:

[0151] An analysis unit, configured to collect images of a specific area based on an AI camera to obtain image data; and analyze the image data based on a structured algorithm in the AI ​​camera to obtain fire safety features;

[0152] A calculation unit, used to obtain current environmental information, input the environmental information into the offset model for calculation, and obtain a corresponding offset value;

[0153] An offset unit, configured to offset a preset parameter of a preset fire safety prediction model based on the offset value to obtain a fire safety prediction model after the offset;

[0154] A prediction unit is used to input the fire safety feature into the offset fire safety prediction model for prediction, obtain a fire safety prediction result, and generate a corresponding fire safety management plan based on the fire safety prediction result.

[0155] In this embodiment, for the specific implementation of each unit in the above system embodiment, please refer to the above method embodiment, which will not be repeated here.

[0156] Reference Figure 3In an embodiment of the present invention, a computer device is also provided. The computer device may be a server, and its internal structure may be as follows: Figure 3 As shown. The computer device includes a processor, a memory, a display screen, an input device, a network interface and a database connected through a system bus. Among them, the processor designed by the computer is used to provide computing and control capabilities. The memory of the computer device includes a non-volatile storage medium and an internal memory. The non-volatile storage medium stores an operating system, a computer program and a database. The internal memory provides an environment for the operation of the operating system and the computer program in the non-volatile storage medium. The database of the computer device is used to store the corresponding data in this embodiment. The network interface of the computer device is used to communicate with an external terminal through a network connection. When the computer program is executed by the processor, the above method is implemented.

[0157] Those skilled in the art will understand that Figure 3 The structure shown in the figure is merely a block diagram of a portion of the structure related to the solution of the present invention, and does not constitute a limitation on the computer device to which the solution of the present invention is applied.

[0158] An embodiment of the present invention further provides a computer-readable storage medium on which a computer program is stored, and when the computer program is executed by a processor, the above method is implemented. It can be understood that the computer-readable storage medium in this embodiment can be a volatile readable storage medium or a non-volatile readable storage medium.

[0159] In summary, the fire safety management method and system based on the structured algorithm of the AI ​​camera in a specific area provided in the embodiment of the present invention include: acquiring images of a specific area based on an AI camera to obtain image data; analyzing the image data based on the structured algorithm in the AI ​​camera to obtain fire safety features; obtaining current environmental information, inputting the environmental information into an offset model for calculation, and obtaining a corresponding offset value; offsetting the preset parameters of a preset fire safety prediction model based on the offset value to obtain a fire safety prediction model after offset; inputting the fire safety features into the fire safety prediction model after offset for prediction to obtain a fire safety prediction result, and generating a corresponding fire safety management plan based on the fire safety prediction result. In the present invention, by inputting environmental information into the offset model for calculation to obtain a corresponding offset value, and offsetting the preset parameters of the preset fire safety prediction model, the offset fire safety prediction model can more accurately predict and analyze the fire safety features, so that the fire safety prediction results are more accurate, thereby generating a reasonable fire safety management plan.

[0160] Those skilled in the art can understand that all or part of the processes in the above-mentioned embodiment methods can be completed by instructing the relevant hardware through a computer program, and the computer program can be stored in a non-volatile computer-readable storage medium. When the computer program is executed, it can include the processes of the embodiments of the above-mentioned methods. Among them, any reference to memory, storage, database or other media provided by the present invention and used in the embodiments may include non-volatile and / or volatile memory. Non-volatile memory may include read-only memory (ROM), programmable ROM (PROM), electrically programmable ROM (EPROM), electrically erasable programmable ROM (EEPROM) or flash memory. Volatile memory may include random access memory (RAM) or external cache memory. As an illustration and not limitation, RAM is available in a variety of forms, such as static RAM (SRAM), dynamic RAM (DRAM), synchronous DRAM (SDRAM), double-speed data rate SDRAM (SSRSDRAM), enhanced SDRAM (ESDRAM), synchronous link (Synchlink) DRAM (SLDRAM), memory bus (Rambus) direct RAM (RDRAM), direct memory bus dynamic RAM (DRDRAM), and memory bus dynamic RAM.

[0161] It should be noted that, in this article, the terms "include", "comprises" or any other variations thereof are intended to cover non-exclusive inclusion, so that a process, device, article or method including a series of elements includes not only those elements, but also includes other elements not explicitly listed, or also includes elements inherent to such process, device, article or method. In the absence of further restrictions, an element defined by the sentence "includes a ..." does not exclude the presence of other identical elements in the process, device, article or method including the element.

[0162] The above description is only a preferred embodiment of the present invention, and does not limit the patent scope of the present invention. Any equivalent structure or equivalent process transformation made by using the contents of the present invention specification and drawings, or directly or indirectly applied in other related technical fields, are also included in the patent protection scope of the present invention.

Claims

1. A fire safety management method based on a structured algorithm of an AI camera in a specific area, characterized in that: The following steps are involved: Capturing images of a specific area based on an AI camera to obtain image data; analyzing the image data based on a structured algorithm in the AI ​​camera to obtain fire safety features; Acquire current environmental information, input the environmental information into the offset model for calculation, and obtain a corresponding offset value; offsetting the preset parameters of the preset fire safety prediction model based on the offset value to obtain the offset fire safety prediction model; Inputting the fire safety feature into the fire safety prediction model after the shift to perform prediction, obtaining a fire safety prediction result, and generating a corresponding fire safety management plan based on the fire safety prediction result; The fire safety characteristics include flame shape characteristics, smoke concentration characteristics, and heat source characteristics; the environmental information includes humidity information, wind information, and ambient temperature information; The image data is analyzed based on the structured algorithm in the AI ​​camera to obtain fire safety features, including: Performing color space conversion on the image data based on the structured algorithm in the AI ​​camera to extract color features of a specified color channel; the color features include flame color features and smoke color features; Performing edge detection on the image data based on the structured algorithm in the AI ​​camera to obtain edge features; the edge features include flame edge features and smoke edge features; Combining the flame color feature and the flame edge feature, a flame shape feature is obtained, and based on the smoke color feature and the smoke edge feature, a smoke concentration feature is obtained; Dividing the image data into different regions based on a structured algorithm in the AI ​​camera, and extracting temperature features of each region as heat source features; The environmental information is input into the offset model for calculation to obtain the corresponding offset value, including: The environmental temperature information, humidity information, and wind force information are respectively input into corresponding sub-offset models for calculation to obtain temperature offset values, humidity offset values, and wind force offset values; wherein each sub-offset model is trained based on the relationship between the corresponding environmental information and fire safety characteristics in historical data; The temperature offset value, humidity offset value and wind force offset value are fused to obtain a final offset value.

2. The fire safety management method based on the specific area AI camera structured algorithm according to claim 1 is characterized in that: Based on the offset value, the preset parameters of the preset fire safety prediction model are offset to obtain the offset fire safety prediction model, including: Based on the offset value, an offset calculation is performed on the connection weights between different neurons in the fire safety prediction model to obtain a fire safety prediction model after offset.

3. The fire safety management method based on the specific area AI camera structured algorithm according to claim 1 is characterized in that: After generating a corresponding fire safety management plan based on the fire safety prediction result, it includes: Acquire multiple fire terminals associated with the specific area; Negotiating an encrypted communication key with each of the fire-fighting terminals; After the fire safety management plan is encrypted based on the encrypted communication key, it is distributed to each of the fire terminals; after each of the fire terminals receives the fire safety management plan, it reminds firefighters to perform fire management.

4. The fire safety management method based on the specific area AI camera structured algorithm according to claim 3 is characterized in that: The step of negotiating an encrypted communication key with each of the fire-fighting terminals comprises: Obtaining the fire management key of each fire terminal; sorting each fire terminal, and combining the fire management keys corresponding to each fire terminal in sequence according to the sorting to obtain a key string; Obtain a binary tree structure template in a preset format, and add the characters in the key string to each node position of the binary tree structure template in sequence to obtain a key binary tree; Obtaining the generation time of the fire safety management plan, and adding the numeric characters in the generation time to the key binary tree in sequence one by one to obtain a changed binary tree; wherein, starting from the root node of the key binary tree, each numeric character is added after the character at each node position to form a character combination; Based on the changed binary tree, the encrypted communication key is generated.

5. The fire safety management method based on the specific area AI camera structured algorithm according to claim 4 is characterized in that: The step of generating the encrypted communication key based on the changed binary tree comprises: Obtain the number corresponding to the number of the fire terminals as the first serial number; remove duplicate characters in the key string to obtain a deduplicated string, and obtain the number corresponding to the number of characters in the deduplicated string as the second serial number; obtain two preset serial numbers; According to a preset order, a sequence number is set for each node position of the changed binary tree; In the changed binary tree, nodes corresponding to the first sequence number, the second sequence number and two preset sequence numbers are obtained as target nodes, and nodes other than the target nodes in the changed binary tree are used as key nodes; Connect each target node in sequence to obtain a plurality of straight lines; generate the encryption communication key according to the positional relationship between each key node and the plurality of straight lines.

6. A fire safety management system based on a specific area AI camera structured algorithm, characterized in that: include: An analysis unit, configured to collect images of a specific area based on an AI camera to obtain image data; and analyze the image data based on a structured algorithm in the AI ​​camera to obtain fire safety features; A calculation unit, used to obtain current environmental information, input the environmental information into the offset model for calculation, and obtain a corresponding offset value; An offset unit, configured to offset a preset parameter of a preset fire safety prediction model based on the offset value to obtain a fire safety prediction model after the offset; A prediction unit, configured to input the fire safety feature into the fire safety prediction model after the offset to perform prediction, obtain a fire safety prediction result, and generate a corresponding fire safety management plan based on the fire safety prediction result; The fire safety characteristics include flame shape characteristics, smoke concentration characteristics, and heat source characteristics; the environmental information includes humidity information, wind information, and ambient temperature information; The image data is analyzed based on the structured algorithm in the AI ​​camera to obtain fire safety features, including: Performing color space conversion on the image data based on the structured algorithm in the AI ​​camera to extract color features of a specified color channel; the color features include flame color features and smoke color features; Performing edge detection on the image data based on the structured algorithm in the AI ​​camera to obtain edge features; the edge features include flame edge features and smoke edge features; Combining the flame color feature and the flame edge feature, a flame shape feature is obtained, and based on the smoke color feature and the smoke edge feature, a smoke concentration feature is obtained; Dividing the image data into different regions based on a structured algorithm in the AI ​​camera, and extracting temperature features of each region as heat source features; The environmental information is input into the offset model for calculation to obtain the corresponding offset value, including: The environmental temperature information, humidity information, and wind force information are respectively input into corresponding sub-offset models for calculation to obtain temperature offset values, humidity offset values, and wind force offset values; wherein each sub-offset model is trained based on the relationship between the corresponding environmental information and fire safety characteristics in historical data; The temperature offset value, humidity offset value and wind force offset value are fused to obtain a final offset value.

7. A computer device comprising a memory and a processor, wherein a computer program is stored in the memory, wherein: When the processor executes the computer program, the steps of the method according to any one of claims 1 to 5 are implemented.

Images