Multi-dimensional fire safety management system

Abstract

The invention relates to the technical field of fire safety, in particular to a multi-dimensional fire safety management system which comprises an environment monitoring module, a data analysis module, a real-time video monitoring module, an intelligent decision module and a self-adaptive evacuation navigation module. The data analysis module processes and analyzes the collected environmental data, the real-time video monitoring module monitors the internal situation of the building by adopting a high-resolution camera, the intelligent decision-making module generates a fire coping strategy, and the self-adaptive evacuation navigation module dynamically generates and updates a safe evacuation route in the building. The self-adaptive evacuation navigation module is introduced, so that a safe evacuation route in the building can be dynamically generated and updated in real time. The system can adjust the evacuation route according to the current environmental condition by analyzing the fire behavior development and the personnel distribution condition in the building in real time, thereby effectively preventing the personnel from entering a high-risk area by mistake, and ensuring that the personnel can be quickly and orderly guided to a safe area in an emergency.

Description

Technical Field

[0001] The present invention relates to the field of fire safety technology, and in particular to a multi-dimensional fire safety management system. Background Art

[0002] With the acceleration of urbanization and the complexity of building structures, traditional fire safety management systems have been unable to meet the needs of quickly and effectively responding to emergency situations such as fires. Especially in high-rise and super high-rise buildings, once a fire occurs, how to quickly and accurately assess the fire situation and guide personnel to evacuate safely becomes the key to ensuring personal safety.

[0003] Traditional fire protection systems mainly rely on passive monitoring devices such as smoke detectors and temperature sensors. Although these devices provide certain early warnings at the beginning of a fire, they have significant deficiencies in data processing, real-time performance, and the formulation of response measures. For example, these systems are usually unable to provide real-time information on the smoke diffusion path and the best evacuation route, resulting in low evacuation efficiency and prone to secondary accidents. Summary of the invention

[0004] Based on the above objectives, the present invention provides a multi-dimensional fire safety management system.

[0005] The multi-dimensional fire safety management system includes an environmental monitoring module, a data analysis module, a real-time video monitoring module, an intelligent decision-making module, and an adaptive evacuation navigation module, among which;

[0006] The environmental monitoring module is responsible for collecting environmental data, including temperature, humidity and smoke concentration;

[0007] The data analysis module processes and analyzes the collected environmental data, monitors and evaluates fire risks in real time, and uses machine learning algorithms to predict fire development trends;

[0008] The real-time video surveillance module uses a high-resolution camera to monitor the interior of the building and provide visual data to support environmental monitoring and emergency response;

[0009] The intelligent decision-making module combines the outputs of the data analysis module and the real-time video monitoring module to generate a fire response strategy, including optimized emergency response measures and evacuation instructions;

[0010] The adaptive evacuation navigation module uses data from the real-time video monitoring module, the data analysis module and the environmental monitoring module to dynamically generate and update safe evacuation routes within the building, adjust the evacuation routes according to the real-time data, and ensure that a safe evacuation path is provided under changing environmental conditions.

[0011] Furthermore, the environmental monitoring module specifically includes:

[0012] Temperature sensor: used to monitor temperature changes in the environment. Abnormal temperature rise may be an early sign of fire.

[0013] Humidity sensor: used to monitor humidity changes in the environment. Rapid changes in humidity are associated with fire, and fire sources can quickly affect humidity levels in the air.

[0014] Smoke detectors: Used to detect smoke particles suspended in the air, which is one of the most common signs of fire.

[0015] Furthermore, the data analysis module specifically includes:

[0016] Data preprocessing: Clean and standardize the collected environmental data, including removing noise, processing missing values, and normalizing the data range;

[0017] Feature engineering: Extract features from environmental data, including the rate of change, historical averages, and peaks of temperature and smoke concentration;

[0018] Real-time monitoring and anomaly detection: Use threshold-based models to monitor whether the data shows unusual patterns;

[0019] Fire trend prediction: Use machine learning algorithms to predict the development trend of fires.

[0020] Furthermore, the real-time monitoring and anomaly detection specifically include:

[0021] Determine thresholds: Thresholds are set based on historical data analysis or industry standards;

[0022] Temperature threshold: Set the safety threshold to 60°C. If the area temperature exceeds this, it indicates a fire risk.

[0023] Smoke concentration threshold: Set the safety threshold to 1%-3% opacity per meter. If the smoke concentration in the air exceeds the preset threshold, it indicates that there is a fire risk.

[0024] Anomaly detection: The data analysis module continuously monitors the collected environmental data. When the environmental data is read, it is compared with the preset threshold. If the environmental data is within the threshold, the environmental status is considered normal. If the environmental data exceeds the threshold, it will be marked as abnormal.

[0025] Furthermore, the fire trend prediction specifically includes:

[0026] Data collection: Collect historical environmental data including temperature, humidity, smoke density, and the time, location, and response measures of the fire;

[0027] Data processing: remove or fill in missing values, standardize the data so that different features have the same scale;

[0028] Feature extraction: Extract features, including temperature change rate and historical smoke concentration;

[0029] Feature engineering: calculate the rolling mean of environmental data, calculate the rolling standard deviation of environmental data, time lag features, use the environmental data of the previous hour or the previous day as features, interactive features of temperature and smoke concentration, interactive features of time and environmental data;

[0030] Construct training and test sets: Divide the data into training and test sets, with a ratio of 80% training set and 20% test set, and use the rolling window technique to create training samples;

[0031] Select kernel function: Select radial basis function kernel;

[0032] Model training: Use support vector machine for training and optimize parameters;

[0033] Model evaluation: The test set is used to evaluate the model performance, and the mean square error and determination coefficient are used as evaluation indicators.

[0034] Furthermore, the real-time video monitoring module specifically includes:

[0035] Real-time video data capture: Install high-resolution cameras at key locations in the building, including entrances and corridors, using digital video technology to ensure that video data is captured and transmitted in real time;

[0036] Real-time analysis of visual data: Use video analysis software to process video streams in real time, including using image processing technology to detect smoke, fire or abnormal personnel in the picture, and applying convolutional neural network models to automatically identify signs of fire in the video.

[0037] Furthermore, the convolutional neural network model specifically includes:

[0038] Data collection: Collect video data containing flames, smoke, and normal scenes. The video data sources include actual surveillance videos, open source datasets, and simulated videos.

[0039] Data labeling: Extract frames from video data and label frames containing flames and smoke. Labeled frames will be used as positive samples when training the convolutional neural network model, while normal frames will be used as negative samples.

[0040] Data preprocessing: Decompose the video into individual frames, preprocess each frame, including resizing, normalizing pixel values ​​to the 0-1 range, and applying data augmentation techniques to increase data diversity, including random rotation, scaling, cropping, and horizontal flipping;

[0041] Build a convolutional neural network model: Design a convolutional neural network model to include multiple convolutional layers, pooling layers, and fully connected layers;

[0042] Model training: Use the labeled frame data to train the convolutional neural network model, and use the cross entropy loss function to measure the difference between the predicted probability distribution of the convolutional neural network model and the actual label.

[0043] Furthermore, the intelligent decision-making module specifically includes:

[0044] Data reception: The intelligent decision-making module receives data from the data analysis module and the real-time video monitoring module;

[0045] Real-time analysis: Identify the severity of the fire and assess the fire risk level based on the data from the data analysis module and the real-time video monitoring module;

[0046] Generate response strategies: Based on the risk assessment results, the intelligent decision-making module uses a preset rule engine to develop response strategies. These rules are based on historical data, previous fire cases and best practices to develop specific response measures for the fire brigade, including the optimal entry points and areas that require special attention, and develop evacuation plans, including determining safe evacuation routes and evacuation destinations.

[0047] Furthermore, the adaptive evacuation navigation module specifically includes:

[0048] Data reception: The adaptive evacuation navigation module receives data from the environmental monitoring module and the data analysis module, and obtains real-time image data inside the building from the real-time video monitoring module;

[0049] Data processing and analysis: Analyze the received data to determine the location, intensity and spread rate of the fire, and identify high-risk areas and relatively safe areas inside the building;

[0050] Route calculation: Use graph theory algorithms to calculate the shortest path from each location to the nearest safe exit on the building's floor plan, taking into account real-time fire conditions and congestion, and dynamically adjust the path to avoid high-risk areas and congested areas.

[0051] Furthermore, the graph theory algorithm adopts Dijkstra algorithm, which specifically includes:

[0052] Initialization: Set the shortest path estimate values ​​of all nodes to infinity, except for the source node, whose path value is set to 0. Use a priority queue to store all nodes and their path values, and the priority queue is sorted by path value.

[0053] Algorithm process: Take the node with the smallest path value from the priority queue, update the path values ​​of all adjacent nodes of the node, and if the path length from the current node to the adjacent node is less than the known shortest path value, update the path value. Repeat the above process until the priority queue is empty. The formula is expressed as:

[0054] dist[v]=min(dist[v],dist[u]+weight(u,v)), where dist[v] is the currently known shortest path length from the source point to node v, dist[u] is the shortest path length from the source point to the current node u, and weight(u,v) is the weight of the edge from node u to v.

[0055] The present invention integrates the environmental monitoring module and the real-time video monitoring module, uses high-resolution cameras and various sensors to collect key data in real time, and combines the advanced machine learning algorithm used in the data analysis module to monitor and accurately assess fire risks in real time. This not only greatly improves the accuracy of fire warnings, but also can quickly identify potential risks in the early stages of a fire, making emergency response measures more timely and effective, significantly reducing casualties and property losses caused by fires.

[0056] The present invention introduces an adaptive evacuation navigation module that can dynamically generate and update safe evacuation routes in a building in real time. By analyzing the development of the fire and the distribution of people in the building in real time, the system can adjust the evacuation route according to the current environmental conditions, effectively preventing people from mistakenly entering high-risk areas. This function not only increases the safety of the evacuation process, but also significantly improves the evacuation efficiency, ensuring that people can be quickly and orderly guided to a safe area in an emergency.

[0057] The present invention allows the system to comprehensively utilize information from multiple data sources through an intelligent decision-making module to develop targeted fire response strategies. This greatly enhances the system's flexibility in responding to complex situations. Whether facing rapidly changing fire conditions or special needs of different buildings, the system can provide reasonable and effective management and response solutions, enhancing the comprehensive response capabilities of the entire fire safety management system. BRIEF DESCRIPTION OF THE DRAWINGS

[0058] In order to more clearly illustrate the technical solutions in the present invention or the prior art, the drawings required for use in the embodiments or the description of the prior art will be briefly introduced below. Obviously, the drawings in the following description are only for the present invention. For ordinary technicians in this field, other drawings can be obtained based on these drawings without paying creative work.

[0059] Figure 1 A schematic diagram of a system flow of an embodiment of the present invention;

[0060] Figure 2 The figure is a schematic diagram of the process of an adaptive evacuation navigation module according to an embodiment of the present invention. DETAILED DESCRIPTION

[0061] In order to make the objectives, technical solutions and advantages of the present invention more clearly understood, the present invention is further described in detail below in conjunction with specific embodiments.

[0062] It should be noted that, unless otherwise defined, the technical terms or scientific terms used in the present invention should be understood by people with ordinary skills in the field to which the present invention belongs. The "first", "second" and similar words used in the present invention do not indicate any order, quantity or importance, but are only used to distinguish different components. "Include" or "comprise" and similar words mean that the elements or objects appearing before the word include the elements or objects listed after the word and their equivalents, without excluding other elements or objects. "Connect" or "connected" and similar words are not limited to physical or mechanical connections, but may include electrical connections, whether direct or indirect. "Up", "down", "left", "right" and the like are only used to indicate relative positional relationships. When the absolute position of the described object changes, the relative positional relationship may also change accordingly.

[0063] like Figure 1-Figure 2 As shown, the multi-dimensional fire safety management system includes an environmental monitoring module, a data analysis module, a real-time video monitoring module, an intelligent decision-making module and an adaptive evacuation navigation module, among which;

[0064] The environmental monitoring module is responsible for collecting environmental data, including temperature, humidity, and smoke concentration;

[0065] The data analysis module processes and analyzes the collected environmental data, monitors and assesses fire risks in real time, and uses machine learning algorithms to predict fire development trends;

[0066] The real-time video surveillance module uses high-resolution cameras to monitor the interior of the building and provide visual data to support environmental monitoring and emergency response;

[0067] The intelligent decision-making module combines the outputs of the data analysis module and the real-time video monitoring module to generate fire response strategies, including optimized emergency response measures and evacuation instructions;

[0068] The adaptive evacuation navigation module uses data from the real-time video surveillance module, data analysis module and environmental monitoring module to dynamically generate and update safe evacuation routes within the building, adjust the evacuation routes according to real-time data, and ensure that a safe evacuation path is provided under changing environmental conditions.

[0069] The environmental monitoring module specifically includes:

[0070] Temperature sensor: used to monitor temperature changes in the environment. Abnormal temperature rise may be an early sign of fire.

[0071] Humidity sensor: used to monitor humidity changes in the environment. Rapid changes in humidity are associated with fire, and fire sources can quickly affect humidity levels in the air.

[0072] Smoke detectors: Used to detect smoke particles suspended in the air, which is one of the most common signs of fire.

[0073] The data analysis module specifically includes:

[0074] Data preprocessing: Clean and standardize the collected environmental data, including removing noise, processing missing values, and normalizing the data range;

[0075] Feature engineering: Extract features from environmental data, including the rate of change, historical averages, and peaks of temperature and smoke concentration;

[0076] Real-time monitoring and anomaly detection: Use threshold-based models to monitor data to see if it shows unusual patterns. For example, if the temperature rises rapidly in a short period of time, this may indicate the occurrence of a fire. This type of detection usually needs to be done in real time to allow for a quick response.

[0077] Fire trend prediction: Use machine learning algorithms to predict the development trend of fires;

[0078] Through real-time data monitoring, analysis and prediction, the data analysis module provides strong support for the entire fire safety management system, ensuring that each component can respond quickly and effectively when needed, thereby improving overall safety and efficiency.

[0079] Real-time monitoring and anomaly detection include:

[0080] Determine thresholds: Thresholds are set based on historical data analysis or industry standards;

[0081] Temperature threshold: Set the safety threshold to 60°C. If the area temperature exceeds this, it indicates a fire risk.

[0082] Smoke concentration threshold: Set the safety threshold to 1%-3% opacity per meter. If the smoke concentration in the air exceeds the preset threshold, it indicates that there is a fire risk.

[0083] Anomaly detection: The data analysis module continuously monitors the collected environmental data. When the environmental data is read, it is compared with the preset threshold. If the environmental data is within the threshold, the environmental status is considered normal. If the environmental data exceeds the threshold, it will be marked as abnormal.

[0084] Fire trend forecasts specifically include:

[0085] Data collection: Collect historical environmental data including temperature, humidity, smoke density, and the time, location, and response measures of the fire;

[0086] Data processing: Remove or fill missing values, standardize the data so that different features have the same scale. The standardization method is z-score standardization, which is calculated as:

[0087]

[0088] Among them, x is the original data, μ is the mean, and σ is the standard deviation;

[0089] Feature extraction: Extract features, including temperature change rate and historical smoke concentration;

[0090] Feature Engineering: Calculate the rolling mean of environmental data. The rolling mean formula is:

[0091] Where N is the rolling window size, x i is the value of the i-th time step, and the rolling standard deviation of the environmental data is calculated. The rolling standard deviation formula is:

[0092] in, is a rolling average, a time-lagged feature, and uses the environmental data of the previous hour or day as a feature. For example, x t-1 Indicates the temperature of the previous hour, the interactive features of temperature and smoke concentration, such as temperature × smoke concentration, and the interactive features of time and environmental data, such as the changes in environmental characteristics at different time points;

[0093] Construct training and test sets: Divide the data into training and test sets, with a ratio of 80% training set and 20% test set. Use the rolling window technique to create training samples. For example, use the data from the past 24 hours to predict the fire risk in the next hour. The rolling window method is as follows:

[0094] X t =[x t-24 , x t-23 , ..., x t-1 ];

[0095] y t =x t ;

[0096] Select kernel function: Select the radial basis function kernel, calculated as:

[0097] K(x i , x j )=exp(-γ||x i -xj || 2 );

[0098] Among them, γ is the kernel function parameter, which controls the width of the radial basis function kernel;

[0099] Model training: Use support vector machine for training and optimize parameters (such as C and γ);

[0100] Objective function:

[0101] Constraints: y i (w T φ(x i )+b)≥1-ξ i ,ξ i ≥0, where φ(x i ) is a nonlinear mapping function that maps the input to a high-dimensional space;

[0102] Model evaluation: The test set is used to evaluate the model performance. The evaluation indicators are mean square error and determination coefficient, which are calculated as follows:

[0103]

[0104] in, is the average of the actual values.

[0105] The real-time video monitoring module specifically includes:

[0106] Real-time video data capture: Install high-resolution cameras at key locations in the building, including entrances and corridors, using digital video technology to ensure that video data is captured and transmitted in real time;

[0107] Real-time analysis of visual data: Use video analysis software to process video streams in real time, including using image processing technology to detect smoke, fire or abnormal personnel in the picture, and applying convolutional neural network models to automatically identify signs of fire in the video.

[0108] The convolutional neural network model specifically includes:

[0109] Data collection: Collect video data containing flames, smoke, and normal scenes. The video data sources include actual surveillance videos, open source datasets, and simulated videos.

[0110] Data labeling: Extract frames from video data and label frames containing flames and smoke. Labeled frames will be used as positive samples when training the convolutional neural network model, while normal frames will be used as negative samples.

[0111] Data preprocessing: Decompose the video into individual frames, preprocess each frame, including resizing, normalizing pixel values ​​to the 0-1 range, and applying data augmentation techniques to increase data diversity, including random rotation, scaling, cropping, and horizontal flipping;

[0112] Construct a convolutional neural network model: Design a convolutional neural network model to include multiple convolutional layers, pooling layers, and fully connected layers. The convolutional layer uses a set of learnable filters to extract features from the input image. Each filter slides on the input data and calculates the dot product of the filter and the input data to generate a feature map, which is mathematically expressed as: F ij =σ(∑ m ∑ n W mn ·X i+m,j+n +b), where F ij is the value of the feature map at position (i, j), W is the weight matrix of the convolution kernel, X is the input data, b is the bias term, σ is the activation function, and the pooling layer is used to reduce the size of the feature map, thereby reducing the computational complexity and overfitting risk. The most commonly used pooling operations are maximum pooling and average pooling, expressed as: P ij =max(X kl ) or P ij =max(X kl ), where P ij is the output after pooling, X kl It is the sub-region of the input feature map within the pooling window. The fully connected layer "flattens" the features in the network and performs the final classification. The output of the fully connected layer is converted into a probability distribution through the softmax function: Among them, y k is the predicted probability of class k, z k is the linear output of the fully connected layer;

[0113] Model training: Use the labeled frame data to train the convolutional neural network model, and use the cross entropy loss function to measure the difference between the predicted probability distribution of the convolutional neural network model and the actual label. The cross entropy loss is defined as:

[0114]

[0115] Where M is the number of categories, y o,c is the true label (0 or 1) of category c in the observed sample o, p o,c is the probability that an observation o is predicted to be of class c.

[0116] The intelligent decision-making module specifically includes:

[0117] Data reception: The intelligent decision-making module receives data from the data analysis module and the real-time video monitoring module;

[0118] Real-time analysis: Identify the severity of the fire and assess the fire risk level based on the data from the data analysis module and the real-time video monitoring module;

[0119] Generate response strategies: Based on the risk assessment results, the intelligent decision-making module uses a preset rule engine to develop response strategies. These rules are based on historical data, previous fire cases and best practices to develop specific response measures for the fire brigade, including the optimal entry points and areas that require special attention, and develop evacuation plans, including determining safe evacuation routes and evacuation destinations.

[0120] The adaptive evacuation navigation module specifically includes:

[0121] Data reception: The adaptive evacuation navigation module receives data from the environmental monitoring module and the data analysis module, and obtains real-time image data inside the building from the real-time video monitoring module;

[0122] Data processing and analysis: Analyze the received data to determine the location, intensity and spread rate of the fire, and identify high-risk areas and relatively safe areas inside the building;

[0123] Route calculation: Use graph theory algorithms to calculate the shortest path from each location to the nearest safe exit on the building's floor plan, taking into account real-time fire conditions and congestion, and dynamically adjust the path to avoid high-risk areas and congested areas.

[0124] The graph theory algorithm uses the Dijkstra algorithm, which includes:

[0125] Initialization: Set the shortest path estimate values ​​of all nodes to infinity, except for the source node, whose path value is set to 0. Use a priority queue to store all nodes and their path values, and the priority queue is sorted by path value.

[0126] Algorithm process: Take the node with the smallest path value from the priority queue, update the path values ​​of all adjacent nodes of the node, and if the path length from the current node to the adjacent node is less than the known shortest path value, update the path value. Repeat the above process until the priority queue is empty. The formula is expressed as:

[0127] dist[v] = min(dist[v], dist[u] + weight(u,v)), where dist[v] is the current known shortest path length from the source to node v, dist[u] is the shortest path length from the source to the current node u, and weight(u,v) is the weight of the edge from node u to v;

[0128] When applied to building evacuation navigation, the Dijkstra algorithm can combine the building's floor plan and real-time environmental data (such as blockages, dangerous areas, etc.) to dynamically calculate and update the optimal evacuation path to each safe exit, which can not only help improve evacuation efficiency but also significantly increase safety.

[0129] Those skilled in the art should understand that the discussion of any of the above embodiments is merely illustrative and is not intended to imply that the scope of the present invention is limited to these examples. Under the concept of the present invention, the technical features in the above embodiments or different embodiments may be combined, the steps may be implemented in any order, and there are many other variations of the different aspects of the present invention as described above, which are not provided in detail for the sake of simplicity.

[0130] The present invention is intended to cover all such substitutions, modifications and variations that fall within the broad scope of the claims. Therefore, any omissions, modifications, equivalent substitutions, improvements, etc. made within the spirit and principles of the present invention should be included in the protection scope of the present invention.

Claims

1. Multi-dimensional fire safety management system, It is characterized in that It includes environmental monitoring module, data analysis module, real-time video monitoring module, intelligent decision-making module and adaptive evacuation navigation module, among which; The environmental monitoring module is responsible for collecting environmental data, including temperature, humidity and smoke concentration; The data analysis module processes and analyzes the collected environmental data, monitors and evaluates fire risks in real time, and uses machine learning algorithms to predict fire development trends; The real-time video surveillance module uses a high-resolution camera to monitor the interior of the building and provide visual data to support environmental monitoring and emergency response; The intelligent decision-making module combines the outputs of the data analysis module and the real-time video monitoring module to generate a fire response strategy, including optimized emergency response measures and evacuation instructions; The adaptive evacuation navigation module uses data from the real-time video monitoring module, the data analysis module and the environmental monitoring module to dynamically generate and update safe evacuation routes within the building, adjust the evacuation routes according to the real-time data, and ensure that a safe evacuation path is provided under changing environmental conditions.

2. The multi-dimensional fire safety management system according to claim 1, It is characterized in that The environmental monitoring module specifically includes: Temperature sensor: used to monitor temperature changes in the environment; Humidity sensor: used to monitor humidity changes in the environment; Smoke detector: Used to detect smoke particles suspended in the air.

3. The multi-dimensional fire safety management system according to claim 1, It is characterized in that The data analysis module specifically includes: Data preprocessing: Clean and standardize the collected environmental data, including removing noise, processing missing values, and normalizing the data range; Feature engineering: Extract features from environmental data, including the rate of change, historical averages, and peaks of temperature and smoke concentration; Real-time monitoring and anomaly detection: Use threshold-based models to monitor whether the data shows unusual patterns; Fire trend prediction: Use machine learning algorithms to predict the development trend of fires.

4. The multi-dimensional fire safety management system according to claim 3, It is characterized in that The real-time monitoring and anomaly detection specifically include: Determine thresholds: Thresholds are set based on historical data analysis or industry standards; Temperature threshold: Set the safety threshold to 60°C. If the area temperature exceeds this, it indicates a fire risk. Smoke concentration threshold: Set the safety threshold to 1%-3% opacity per meter. If the smoke concentration in the air exceeds the preset threshold, it indicates that there is a fire risk. Anomaly detection: The data analysis module continuously monitors the collected environmental data. When the environmental data is read, it is compared with the preset threshold. If the environmental data is within the threshold, the environmental status is considered normal. If the environmental data exceeds the threshold, it will be marked as abnormal.

5. The multi-dimensional fire safety management system according to claim 3, It is characterized in that The fire trend prediction specifically includes: Data collection: Collect historical environmental data including temperature, humidity, smoke density, and the time, location, and response measures of the fire; Data processing: remove or fill in missing values, standardize the data so that different features have the same scale; Feature extraction: Extract features, including temperature change rate and historical smoke concentration; Feature engineering: calculate the rolling mean of environmental data, calculate the rolling standard deviation of environmental data, time lag features, use the environmental data of the previous hour or the previous day as features, interactive features of temperature and smoke concentration, interactive features of time and environmental data; Construct training and test sets: Divide the data into training and test sets, with a ratio of 80% training set and 20% test set, and use the rolling window technique to create training samples; Select kernel function: Select radial basis function kernel; Model training: Use support vector machine for training and optimize parameters; Model evaluation: The test set is used to evaluate the model performance, and the mean square error and determination coefficient are used as evaluation indicators.

6. The multi-dimensional fire safety management system according to claim 1, It is characterized in that The real-time video monitoring module specifically includes: Real-time video data capture: Install high-resolution cameras at key locations in the building, including entrances and corridors, using digital video technology to ensure that video data is captured and transmitted in real time; Real-time analysis of visual data: Use video analysis software to process video streams in real time, including using image processing technology to detect smoke, fire or abnormal personnel in the picture, and applying convolutional neural network models to automatically identify signs of fire in the video.

7. The multi-dimensional fire safety management system according to claim 6, It is characterized in that The convolutional neural network model specifically includes: Data collection: Collect video data containing flames, smoke, and normal scenes. The video data sources include actual surveillance videos, open source datasets, and simulated videos. Data labeling: Extract frames from video data and label frames containing flames and smoke. Labeled frames will be used as positive samples when training the convolutional neural network model, while normal frames will be used as negative samples. Data preprocessing: Decompose the video into individual frames, preprocess each frame, including resizing, normalizing pixel values ​​to the 0-1 range, and applying data augmentation techniques to increase data diversity, including random rotation, scaling, cropping, and horizontal flipping; Build a convolutional neural network model: Design a convolutional neural network model to include multiple convolutional layers, pooling layers, and fully connected layers; Model training: Use the labeled frame data to train the convolutional neural network model, and use the cross entropy loss function to measure the difference between the predicted probability distribution of the convolutional neural network model and the actual label.

8. The multi-dimensional fire safety management system according to claim 1, It is characterized in that The intelligent decision-making module specifically includes: Data reception: The intelligent decision-making module receives data from the data analysis module and the real-time video monitoring module; Real-time analysis: Identify the severity of the fire and assess the fire risk level based on the data from the data analysis module and the real-time video monitoring module; Generate response strategies: Based on the risk assessment results, the intelligent decision-making module uses a preset rule engine to develop response strategies. These rules are based on historical data, previous fire cases and best practices to develop specific response measures for the fire brigade, including the optimal entry points and areas that require special attention, and develop evacuation plans, including determining safe evacuation routes and evacuation destinations.

9. The multi-dimensional fire safety management system according to claim 1, It is characterized in that The adaptive evacuation navigation module specifically includes: Data reception: The adaptive evacuation navigation module receives data from the environmental monitoring module and the data analysis module, and obtains real-time image data inside the building from the real-time video monitoring module; Data processing and analysis: Analyze the received data to determine the location, intensity and spread rate of the fire, and identify high-risk areas and relatively safe areas inside the building; Route calculation: Use graph theory algorithms to calculate the shortest path from each location to the nearest safe exit on the building's floor plan, taking into account real-time fire conditions and congestion, and dynamically adjust the path to avoid high-risk areas and congested areas.

10. The multi-dimensional fire safety management system according to claim 9, It is characterized in that The graph theory algorithm adopts Dijkstra algorithm, which specifically includes: Initialization: Set the shortest path estimate values ​​of all nodes to infinity, except for the source node, whose path value is set to 0. Use a priority queue to store all nodes and their path values, and the priority queue is sorted by path value. Algorithm process: Take the node with the smallest path value from the priority queue, update the path values ​​of all adjacent nodes of the node, and if the path length from the current node to the adjacent node is less than the known shortest path value, update the path value. Repeat the above process until the priority queue is empty. The formula is expressed as: dist[v]=min(dist[v], dist[u]+weight(u,v)), where dist[v] is the currently known shortest path length from the source point to node v, dist[u] is the shortest path length from the source point to the current node u, and weight(u,v) is the weight of the edge from node u to v.

Images