Smart campus multi-dimensional fire safety monitoring method and system

Abstract

The invention relates to the technical field of fire safety, and provides a smart campus multi-dimensional fire safety monitoring method and system. The method comprises the following steps: collecting multi-source security data of a campus building; extracting features based on a preset risk dimension, and generating a fire risk feature vector; constructing a risk model according to historical fire cases, and scoring the feature vectors to obtain a real-time risk score; when the score exceeds a dynamic threshold value, triggering a grading early warning instruction; an emergency scheme is generated according to the early warning, and fire fighting equipment and a personnel evacuation system are linked to execute operation. The technical problem that an existing campus fire safety monitoring system lacks a multi-dimensional, real-time and comprehensive risk assessment means and is difficult to accurately judge different fire risks and deal with the fire risks in real time is solved, and accurate fire risk early warning and efficient emergency response are achieved through multi-source data fusion and intelligent analysis. And the campus fire safety management level is improved.

Description

Technical Field

[0001] This application relates to the technical field of fire safety, and particularly to a multi-dimensional fire safety monitoring method and system for smart campuses. Background Art

[0002] With the continuous advancement of smart campus construction, the building structures on campus have become increasingly complex, the personnel mobility has increased, and the functional areas have become diversified. Fire safety issues are facing new challenges. Traditional campus fire safety management methods mostly rely on manual inspections and the deployment of fixed sensors, lacking the comprehensive perception ability of multi-source heterogeneous data, and it is difficult to identify potential fire risks in a timely and accurate manner. In practical applications, factors such as uneven building fire resistance performance, opaque operation status of fire-fighting equipment, drastic fluctuations in environmental parameters, and dense personnel distribution may all become the inducements of fire hazards. However, existing technologies generally lack a comprehensive risk analysis mechanism based on building structure characteristics, equipment status, environmental changes, and personnel dynamics, making it difficult to achieve real-time assessment and dynamic response to fire risks, resulting in delayed warnings and low emergency response efficiency. To improve the overall fire safety level of the campus, it is urgent to introduce intelligent, multi-dimensional, and dynamically adaptable monitoring methods and systems, integrate multi-source data analysis, machine learning modeling, and emergency linkage control technologies, and achieve accurate identification, scientific warning, and efficient disposal of campus fire risks. Summary of the Invention

[0003] This application provides a multi-dimensional fire safety monitoring method and system for smart campuses, aiming to solve the technical problems that the existing campus fire safety monitoring system lacks multi-dimensional, real-time, and comprehensive risk assessment means and is difficult to accurately judge different fire risks and respond in real time.

[0004] In the first aspect disclosed in this application, a multi-dimensional fire safety monitoring method for smart campuses is provided. The method includes: collecting and obtaining multi-source safety data of campus buildings, where the multi-source safety data includes building structure data, fire-fighting equipment status data, environmental sensor data, and personnel distribution data; extracting features from the multi-source safety data based on preset risk dimensions to generate a fire risk feature vector; constructing a multi-dimensional risk probability model according to historical fire case data, calculating a risk score for the fire risk feature vector to obtain a real-time risk score; when the real-time risk score exceeds the dynamic risk threshold, triggering a hierarchical warning instruction; generating an emergency treatment plan according to the hierarchical warning instruction, and linking the fire-fighting equipment control system and the personnel evacuation guidance system to execute emergency operations.

[0005] Another aspect disclosed in this application provides a multi-dimensional fire safety monitoring system for smart campuses. The system includes: a data collection module that collects and obtains multi-source safety data of campus buildings, where the multi-source safety data includes building structure data, fire equipment status data, environmental sensor data, and personnel distribution data; a feature extraction module that extracts features from the multi-source safety data based on preset risk dimensions to generate a fire risk feature vector; a scoring calculation module that constructs a multi-dimensional risk probability model according to historical fire case data, calculates the risk score for the fire risk feature vector, and obtains a real-time risk score; an instruction trigger module that triggers a hierarchical warning instruction when the real-time risk score exceeds the dynamic risk threshold; and an emergency operation module that generates an emergency treatment plan according to the hierarchical warning instruction, and links the fire equipment control system and the personnel evacuation guidance system to execute emergency operations.

[0006] One or more technical solutions provided in this application have at least the following technical effects or advantages:

[0007] The above-mentioned multi-dimensional fire safety monitoring method for smart campuses collects various safety data in campus buildings, including building structure, fire equipment operation status, environmental sensor information, and personnel distribution, to comprehensively perceive the safety situation inside the campus. Subsequently, based on the preset risk dimensions, in-depth analysis and feature extraction are performed on these data to construct a feature vector that can reflect the current fire risk status. Then, combined with a large number of historical fire cases, a multi-dimensional risk probability model is established to score the current feature vector and obtain the real-time fire risk level. When the risk score exceeds the dynamic threshold set by the system, the system will automatically trigger warning instructions of different levels. After that, corresponding emergency treatment plans are generated according to the warning levels, and the fire equipment and the personnel evacuation guidance system are linked to achieve rapid response and intelligent disposal in the initial stage of a fire. This method realizes closed-loop management from risk identification to emergency response and improves the intelligent level of campus fire safety.

[0008] The above description is only an overview of the technical solutions of this application. In order to be able to understand the technical means of this application more clearly, it can be implemented according to the content of the description. And in order to make the above and other purposes, features, and advantages of this application more obvious and understandable, the following specific embodiments of this application are specifically given. Description of the Drawings

[0009] In order to more clearly illustrate the technical solutions in the embodiments of the present invention, the following will briefly introduce the drawings required for the description of the embodiments. Obviously, the following drawings are only some embodiments of the present invention. For those of ordinary skill in the art, other drawings can be obtained based on these drawings without creative efforts.

[0010] Figure 1 It is a schematic flowchart of a multi - dimensional fire safety monitoring method for a smart campus in an embodiment.

[0011] Figure 2 It is an architecture diagram of a multi - dimensional fire safety monitoring system for a smart campus in an embodiment.

[0012] Explanation of reference numerals: Data acquisition module 11, feature extraction module 12, scoring calculation module 13, instruction trigger module 14, emergency operation module 15. Detailed implementation manners

[0013] In the embodiments of the present application, by providing a multi - dimensional fire safety monitoring method and system for a smart campus, the technical problem that the existing campus fire safety monitoring system lacks multi - dimensional, real - time and comprehensive risk assessment means and is difficult to accurately judge different fire risks and respond in real time is solved.

[0014] Next, the technical solutions in the embodiments of the present application will be clearly and completely described in conjunction with the accompanying drawings in the embodiments of the present application. Obviously, the described embodiments are only a part of the embodiments of the present application, rather than all the embodiments. All other embodiments obtained by those of ordinary skill in the art based on the embodiments of the present application without creative efforts shall fall within the protection scope of the present application.

[0015] It should be noted that the terms "include" and "have" and any variations thereof are intended to cover non - exclusive inclusion. For example, a process, method, system, product or server that includes a series of steps or units does not necessarily limit to those steps or units clearly listed, but may include other steps or modules that are not clearly listed or are inherent to these processes, methods, products or devices.

[0016] Embodiment 1, as Figure 1 shown, the present application provides a multi - dimensional fire safety monitoring method for a smart campus, and the method includes:

[0017] Collect and obtain multi - source safety data of campus buildings, where the multi - source safety data includes building structure data, fire - fighting equipment status data, environmental sensor data, and personnel distribution data.

[0018] In the embodiments of the present application, various types of safety information of campus buildings are collected in real time through multiple sensors and data sources. These data include building structure data (such as the fire resistance rating of buildings, structural design, distribution of safety exits, etc.), fire-fighting equipment status data (such as the pressure of fire extinguishers, working status of sprinkler systems, operation of smoke alarms, etc.), environmental sensor data (such as detection data of temperature, humidity, smoke concentration, carbon dioxide and other gases), and personnel distribution data (such as the number of people on different floors and in different areas, personnel flow paths, personnel density, etc.). These multi-source data provide a basis for comprehensively evaluating the fire risk and can help the system more accurately perceive and analyze the fire safety status inside the campus.

[0019] Feature extraction is performed on the multi-source safety data based on preset risk dimensions to generate a fire risk feature vector.

[0020] In one embodiment, the multi-source safety data collected is analyzed and processed according to preset risk dimensions. First, data such as building structure, fire-fighting equipment, environmental sensors, and personnel distribution are classified according to the preset risk dimensions. Each type of data will be extracted according to the features related to the fire risk. For example, the fire resistance rating of the building, the working status of the equipment, the temperature and smoke concentration in the environment, and the distribution density of personnel are all converted into digital feature values. For non-numerical type data, one-hot encoding can be used for quantization. Subsequently, these feature values are integrated to form a comprehensive fire risk feature vector, which contains all the factors that may affect the occurrence and spread of a fire. In this way, the fire risk status in the current campus can be quantified and represented, providing basic data for subsequent risk assessment and decision-making.

[0021] Furthermore, the present application provides a method for performing feature extraction on the multi-source safety data based on preset risk dimensions, and the method includes:

[0022] The building structure data is divided into a building risk dimension, and the fire resistance rating of the building, the safety exit density, and the fire compartment integrity index are extracted; the fire-fighting equipment status data is divided into an equipment availability dimension, and the equipment failure rate, the pressure value of the fire extinguisher, and the response time of the smoke detector are extracted; the environmental sensor data is divided into an environmental hazard dimension, and the temperature change rate, the smoke concentration gradient, and the peak CO concentration are extracted; the personnel distribution data is divided into a personnel density dimension, and the number of people on the floor, the congestion index of the escape route, and the number of people staying in key areas are extracted; the data of the building risk dimension, the equipment availability dimension, the environmental hazard dimension, and the personnel density dimension are normalized to generate the fire risk feature vector.

[0023] Preferably, the preset risk dimensions are analyzed to determine the required dimension data types, and specific features related to fire risks are extracted from multi-source security data according to these dimension data types. Specifically, building risk dimension features are extracted from the building structure data of multi-source security data to obtain the fire resistance rating of the building, the safety exit density, and the fire compartment integrity index. Among them, the fire resistance rating of the building is used to evaluate the fire resistance ability of building materials and structures. The fire resistance rating of the building is divided into A, B, C, etc., indicating the fire resistance ability of building materials. The safety exit density is the ratio of the number of available safety exits in each floor of the building to the total building area, which is used to reflect the emergency evacuation ability. The fire compartment integrity index is used to evaluate the integrity of each fire compartment in the building, such as whether the firewalls are intact and the sealing performance of doors and windows. Equipment availability dimension features are extracted from the fire equipment status data of multi-source security data to obtain the equipment failure rate, the pressure value of the fire extinguisher, and the response time of the smoke detector. Among them, the equipment failure rate is used to reflect the reliability of the equipment. The pressure value of the fire extinguisher is used to determine whether the pressure of the fire extinguisher is within the normal range. Too low a pressure value may indicate that the fire extinguisher has failed. The response time of the smoke detector is used to reflect the sensitivity and response speed of the equipment. Environmental hazard dimension features are extracted from the environmental sensor data of multi-source security data to obtain the temperature change rate, the smoke concentration gradient, and the CO concentration peak value. Among them, the temperature change rate is the rate of temperature change in the environment. A rapid increase in temperature may be a sign of fire spread. The smoke concentration gradient is used to reflect the difference in smoke concentration in different areas. Areas with a large gradient may be the fire source or the fire spread area. The CO concentration peak value is the maximum value of the carbon monoxide concentration in the air. An increase in the carbon monoxide concentration is usually one of the indicators of a fire. Personnel density dimension features are extracted from the personnel distribution data of multi-source security data to obtain the number of people on each floor, the congestion index of the escape route, and the number of people staying in key areas. Among them, the number of people on each floor is the real-time number of people on each floor. The congestion index of the escape route is the degree of crowding of people in the escape route. An overly congested route will reduce the evacuation efficiency. The number of people staying in key areas is the number of people staying in key areas (such as laboratories, classrooms, etc.). Too many people staying may affect the evacuation efficiency. After obtaining the characteristic data of these dimensions, the data of each dimension will be normalized so that different types of data are converted into standard comparable values. Common normalization methods include maximum normalization, Z-score standardization, etc. Through this step, it is ensured that data of different dimensions are compared on the same scale, and features of different dimensions can be directly combined. Finally, the normalized characteristic values of each dimension are spliced to form a comprehensive fire risk characteristic vector. This characteristic vector integrates risk information in various aspects such as building structure, fire equipment status, environmental hazard, and personnel density, and can comprehensively reflect the comprehensive state of fire risks on campus, serving as the basis for subsequent risk assessment and treatment decisions.

[0024] Construct a multi-dimensional risk probability model based on historical fire case data, calculate the risk score for the fire risk feature vector, and obtain the real-time risk score.

[0025] In one embodiment, historical fire cases of campuses or similar buildings are collected and sorted out via networking, including information such as the time and location of the fire, the fire development process, the loss situation, and the fire response. By feature marking and recording each of these historical fire cases, key data such as the building structure characteristics, equipment status, environmental parameters, and personnel distribution of each historical fire case can be obtained. Subsequently, based on this data, a building risk detection unit, an equipment risk detection unit, an environmental risk detection unit, and a personnel risk detection unit are respectively constructed using the method of supervised learning. These units can perform score prediction based on the received data and obtain a comprehensive risk score through weighting. By integrating these units, a multi-dimensional risk probability model can be obtained. When the multi-dimensional risk probability model receives the fire risk feature vector, it disassembles the fire risk feature vector to obtain multiple sub-vectors, and each sub-vector corresponds to a type of data, such as a building structure vector, an equipment status vector, etc. Then, the multi-dimensional risk probability model uses multiple internal units to calculate the risk scores for these sub-vectors, thereby predicting multiple scores. After weighted summation calculation by the model output layer, the real-time risk score is obtained. This score reflects the level of fire risk in the current building environment. The higher the score, the greater the fire risk. This real-time score is a dynamic indicator and will be updated as the real-time data changes. The system continuously updates the risk score according to various types of real-time collected data (such as environmental temperature, smoke concentration, equipment status, etc.), providing decision support for subsequent early warning and emergency response.

[0026] Furthermore, the present application provides that the multi-dimensional risk probability model is constructed in the following manner:

[0027] Obtain historical fire case data, and mark the corresponding building structure features, equipment status features, environmental features, and personnel features for each event; use the building structure features and their building structure scores as inputs to train the building risk detection unit; use the equipment status features and their equipment status scores as inputs to train the equipment risk detection unit; use the environmental features and their environmental scores as inputs to train the environmental risk detection unit; use the personnel features and their personnel scores as inputs to train the personnel risk detection unit; integrate the building risk detection unit, the equipment risk detection unit, the environmental risk detection unit, and the personnel risk detection unit to generate the multi-dimensional risk probability model.

[0028] Preferably, collect the fire case data of campuses or similar buildings that have occurred in history, and mark the building structure characteristics, equipment status characteristics, environmental characteristics, and personnel characteristics in these case data. These characteristics include corresponding dimensional data (such as building fire resistance rating, equipment failure rate, temperature change rate, etc.) and corresponding characteristic scores. For example, for building structure characteristics, it includes building fire resistance rating, safety exit density, fire compartment integrity index, and building structure score. Subsequently, use the building structure characteristics and their corresponding building structure scores as inputs, and use supervised learning algorithms (such as decision trees, support vector machines, neural networks, etc.) to train the building risk detection unit. This unit will be able to evaluate the risk status of the building based on the input building structure characteristics and predict whether the building structure is likely to exacerbate the spread of fire or cause greater damage during a fire. Similarly, use the equipment status characteristics and their corresponding equipment status scores as inputs to train the equipment risk detection unit. This unit evaluates whether the fire-fighting equipment is in good condition based on the working status of the equipment, such as whether the fire extinguisher is effective and whether the smoke detector can respond in a timely manner, so as to evaluate whether the equipment can effectively cope with fire risks. Similarly, use the environmental characteristics and their corresponding environmental scores as inputs to train the environmental risk detection unit. This unit evaluates the possibility of a fire by analyzing environmental conditions (such as temperature, humidity, gas concentration, etc.). For example, changes in high temperature, thick smoke, or CO concentration may be precursors to a fire, and the environmental risk unit can predict the environmental risk status based on these factors. Similarly, use the personnel characteristics and their corresponding personnel scores as inputs to train the personnel risk detection unit. This unit evaluates the impact of personnel distribution on the fire, including the evacuation ability of personnel during a fire, passage congestion, the number of people staying in key areas, etc. The density of personnel and the smoothness of the evacuation path have a huge impact on the fire. This unit can evaluate the difficulty of personnel evacuation and potential risks based on these factors. Finally, parallel integrate the building risk detection unit, equipment risk detection unit, environmental risk detection unit, and personnel risk detection unit to form a comprehensive multi-dimensional risk probability model. This model can comprehensively evaluate the severity of a fire based on multiple factors such as the input building structure, equipment status, environmental data, and personnel distribution, and generate a real-time risk score based on the evaluation results. Through this training process, it is possible to accurately evaluate the current fire risk based on historical fire data, real-time data input, and intelligent models, and conduct risk prediction and management based on the multi-dimensional evaluation results.

[0029] Furthermore, the present application provides a method for training a building risk detection unit, including:

[0030] The building structure features are paired with the corresponding building structure scores to construct a training sample set; the training sample set is subjected to feature normalization and label standardization to generate standardized training data; the standardized training data is input into an initial building risk detection unit and trained using a supervised learning algorithm to form a building risk detection unit capable of evaluating building structure risks.

[0031] Optionally, building structural features are extracted from historical fire case data. These features may include the fire resistance level of the building, the number of safety exits, the integrity of the fire partition, the building structure score, etc. A set of training sample sets is constructed by pairing the fire resistance level, the number of safety exits, and the integrity of the fire partition with the building structure score of each building. Each sample consists of a building structure feature and a corresponding score to form an input-output pair. Subsequently, since different building structure features (such as fire resistance level, number of safety exits, etc.) may have different dimensions or ranges, these features need to be normalized. Common methods include maximum value normalization (scaling each feature to between 0 and 1) or Z-score standardization (converting the data to a standard normal distribution with a mean of 0 and a variance of 1). By normalizing these data and then splicing the processed data, multiple sample feature vectors can be obtained, and the building structure score is used as the label of each sample feature vector, thereby generating standardized training data. Through normalization and standardization, all features and labels will be unified in the same dimension and range to avoid excessive influence of different dimensional features on model training. After that, the standardized training data set is input into the initial building risk detection unit, which is a model built based on machine learning algorithms (such as support vector machines, decision trees, random forests, neural networks, etc.). Taking the multi-layer perceptron (MLP) as an example, a building risk detection unit is constructed using MLP, including an input layer, a hidden layer, an output layer, etc. The weight of the building risk detection unit is initialized using random numbers, etc., and the standardized training data is input into the initialized building risk detection unit for forward propagation. The features in the building structure data are extracted through the input layer, hidden layer, output layer, etc., and the risk score of the building structure is generated. Then, the mean square error loss function is used to calculate the loss value between the predicted result and the actual label, and the gradient of the loss to the weight of each layer is calculated layer by layer through back propagation. The Adam optimizer is then used to optimize the unit parameters and adjust the weights to minimize the value of the loss function. The above process is repeated until the maximum number of iterations is reached. After the training is completed, the unit performance is tested using the validation set to evaluate the accuracy of the unit in the building structure risk assessment task. If the accuracy meets expectations, the current building risk detection unit is output. Otherwise, hyperparameters such as learning rate and number of training batches are adjusted to further improve the effect of the unit in building structure risk assessment.

[0032] When the real-time risk score exceeds the dynamic risk threshold, a hierarchical warning instruction is triggered.

[0033] In one embodiment, when the calculated real-time fire risk score exceeds the preset dynamic risk threshold, it means that the current fire risk has reached a certain dangerous level and may pose a threat to campus safety. At this time, the hierarchical warning mechanism will be automatically triggered, and corresponding warning instructions will be issued according to different risk levels (such as yellow, orange, and red warnings). Each warning level corresponds to different response measures to help relevant personnel take preventive or emergency measures in a timely manner to reduce the potential fire risk. Through this hierarchical warning instruction, different situations can be dynamically responded to, ensuring the fire safety of the campus.

[0034] Furthermore, the present application provides that the determination of the dynamic risk threshold includes:

[0035] Obtain the campus activity scene type of the current period, and the scene type includes class time, assembly time, and night dormancy time; match the preset basic threshold according to the scene type, wherein the basic threshold corresponds to the class time, the assembly time dynamically adjusts the basic threshold based on the risk of personnel gathering, and the night dormancy time dynamically adjusts the basic threshold based on the risk of response delay; superimpose the impact assessment of the real-time weather conditions, and when it is monitored that the meteorological environment exacerbates the fire spread risk, adaptively adjust the basic threshold to obtain the dynamic risk threshold.

[0036] Preferably, first identify and obtain the type of campus activity scenario in the current period. According to different time periods, there are differences in the activities and behaviors of people on campus. Common scenario types include class periods, assembly periods, and nighttime dormancy periods. Among them, class periods are generally during the day, and people are relatively dispersed in various areas such as classrooms and laboratories, with relatively low evacuation difficulty; assembly periods are like campus activities, lectures, or assemblies, where people are concentrated in a certain area, resulting in greater evacuation pressure; during the nighttime dormancy period, most people are in dormitories or rest areas, with less personnel flow, and usually a longer response delay after a fire occurs. Subsequently, match the preset basic threshold according to the scenario type. For class periods, due to the dispersion of people and the normal use of building facilities, the fire risk at this time is relatively low, so a lower basic threshold is set to reflect the fire risk level of this period. For assembly periods, considering the concentration of people, the difficulty of evacuating people and potential risks are greater when a fire occurs. At this time, the basic threshold will be lowered to improve the warning sensitivity, and the lowering range is generally 20%, which can be adjusted according to actual needs. For the nighttime dormancy period, since people are in a resting state and the response delay after a fire occurs is long, more sensitive warnings are required. At this time, the basic threshold will be further lowered, generally by 30%. After that, monitor the weather changes in real time, especially meteorological conditions such as temperature, humidity, and wind speed. When the meteorological conditions exacerbate the fire spread risk, that is, when one or more of the temperature, humidity, and wind speed exceed the meteorological impact threshold, the basic threshold will be adaptively adjusted according to the weather factors. For example, a high wind speed or high temperature may accelerate the spread of the fire and increase the fire risk. At this time, calculate the difference between these abnormal parameters and the threshold, calculate the ratio with the threshold, and then perform weighted summation on the calculation results to obtain the influence coefficient. Then, multiply the influence coefficient by the basic threshold, and subtract the obtained product from the basic threshold to obtain a dynamic risk threshold. This dynamic risk threshold reflects the real-time fire risk level, provides a more accurate reference for subsequent warnings and emergency responses, ensures timely and reasonable responses in different scenarios and environments, and maximally guarantees campus safety.

[0037] Furthermore, the triggering of the hierarchical warning instruction provided by this application includes:

[0038] Set three levels of dynamic risk thresholds for yellow warning, orange warning, and red warning; when the real-time risk score first exceeds the yellow warning threshold, start equipment self-check and warning broadcast; when the score continuously exceeds the orange warning threshold for a preset duration, trigger the pre-start of the sprinkler system and unlock the access control of the escape route; when the score instantaneously exceeds the red warning threshold, execute power-off protection for the whole building and activate emergency lighting and evacuation guidance.

[0039] Optionally, set the three-level dynamic risk thresholds for yellow, orange, and red warnings according to historical experience and expert decisions. Among them, the yellow warning threshold indicates a relatively low risk, but there are potential fire hazards. When the real-time risk score first exceeds the yellow warning threshold, it is considered that the fire risk has entered the warning stage and preliminary responses are required; the orange warning threshold indicates a relatively high fire risk. If the score continues to exceed this threshold, the risk of fire occurrence increases. At this time, more preventive and emergency measures need to be taken; the red warning threshold indicates an extremely high fire risk and there is a serious threat of a fire outbreak. At this time, the largest-scale emergency response should be immediately taken to ensure the safety of personnel and property protection. The system first compares the real-time fire risk score with the yellow warning threshold. Once the score first exceeds the yellow warning threshold, it indicates that the fire risk has increased and preliminary preventive measures need to be initiated. At this time, the device self-check program is automatically started to check the operating status of relevant fire-fighting equipment (such as fire extinguishers, smoke alarms, sprinkler systems, etc.) to ensure that it is in normal working condition. At the same time, the warning broadcast is started to broadcast the fire warning information to all personnel on campus, reminding everyone to pay attention to the fire hazards and get prepared. When the fire risk score continues to be higher than the orange warning threshold and exceeds the set time threshold (such as continuously exceeding 1 minute), it is considered that the fire risk has increased and entered a higher-level emergency response stage. At this time, the sprinkler system is pre-activated, and some valves of the sprinkler system are started in advance so that fire extinguishing can be carried out quickly when a fire occurs. At the same time, the access control of the escape route will be unlocked to ensure that personnel can be evacuated smoothly and avoid personnel being trapped in the building. When the real-time fire risk score instantaneously exceeds the red warning threshold, it indicates that the fire has entered an extremely dangerous state and the largest-scale emergency measures must be taken immediately. At this time, the power supply of the entire building is automatically cut off to protect against the spread of fire caused by electrical equipment or the aggravation of the fire disaster. At the same time, the emergency lighting is activated to ensure that after the power failure, areas such as emergency exits, stairs, and exits in the building remain bright to avoid personnel getting lost during the evacuation process. In addition, the evacuation guidance system will be started to guide personnel to safe areas through electronic displays or broadcasts, etc., to ensure that the evacuation process proceeds in an orderly manner. Through the above process, the emergency response measures can be dynamically adjusted according to the real-time changes of the fire risk, and the corresponding fire-fighting equipment and personnel evacuation mechanisms can be automatically triggered according to different levels of warnings to maximize the safety of campus personnel.

[0040] Generate an emergency treatment plan according to the classified warning instructions, and link the fire equipment control system and the personnel evacuation guidance system to perform emergency operations.

[0041] In one embodiment, according to the real-time fire risk score and warning level, the corresponding emergency treatment plan is automatically selected. For example, a yellow warning may only require starting equipment self-check and warning broadcasts, while a red warning requires a comprehensive emergency response, involving power-off protection, activation of the sprinkler system, etc. The emergency treatment plan not only includes the control of fire-fighting equipment but also the specific arrangements for personnel evacuation. For example, under an orange warning, the sprinkler system may need to be started in advance, and under a red warning, it is ensured that all safety exits are unlocked and evacuation passages are unobstructed. According to the emergency treatment plan, the fire-fighting equipment control system will be linked to automatically start relevant fire-fighting equipment for fire prevention and suppression. For example, fire-fighting equipment such as sprinkler systems, fire extinguishers, and smoke alarms will be automatically activated at different warning levels. Under a red warning, the power-off protection for the entire building will be fully activated to cut off the power to prevent the spread of electrical fires. The sprinkler system will automatically adjust the spraying area according to the spread of the fire and quickly extinguish the fire. At the same time, the personnel evacuation guidance system will be linked to start the emergency evacuation procedure. According to the fire risk assessment, all escape routes will be automatically opened, safety exits will be unlocked, and personnel will be guided to quickly evacuate through emergency broadcasts or signs. When a red warning is issued, emergency lighting will be activated to ensure that the evacuation paths inside the building are still clearly visible in the event of a power outage, preventing personnel from getting lost due to insufficient lighting. The evacuation guidance information will also be updated in real time through electronic displays and voice broadcasts to ensure the orderly evacuation of personnel to a safe area. Through these linked operations, various emergency measures can be quickly and effectively activated in the event of a fire, ensuring the rapid evacuation of personnel and the timely response of fire-fighting equipment, thereby minimizing the losses caused by the fire and protecting the lives and safety of teachers and students.

[0042] Further, the present application provides an emergency treatment plan generated according to the hierarchical warning instruction, including:

[0043] Obtain personnel location data and fire spread data in real time; obtain a three-dimensional evacuation path based on the building BIM model and fire spread data; use the Dijkstra algorithm to calculate the optimal evacuation path based on the personnel location data and three-dimensional evacuation path, and dynamically generate the evacuation guidance.

[0044] Optionally, through sensors, cameras or other positioning devices installed in the building, the location information of each person in the building is tracked and obtained in real time, and this data can be transmitted to the system through wireless networks or other positioning technologies. The spread of the fire is obtained through environmental sensors (such as smoke detectors, temperature sensors, etc.). This data reflects the location of the fire, the spread speed and the areas that may be affected, helping to understand which areas have been occupied by the fire and which areas are still safe. Subsequently, a building BIM model is constructed. This building BIM model contains the structural information of the building, floor plans, locations of key facilities such as emergency exits, evacuation routes, and stairwells. Through the BIM model, the layout of each floor and the building can be clearly understood. Combining with the data on the spread of the fire, it is possible to update in real time which areas are no longer suitable for passage and which areas are still safe. Based on the building BIM model and the data on the spread of the fire, a three-dimensional evacuation path is calculated. This path not only takes into account the physical structure of the building but also avoids the areas affected by the fire, ensuring that people can avoid the fire source and evacuate smoothly to a safe area. Then, the Dijkstra algorithm is activated. The Dijkstra algorithm is an algorithm used to calculate the shortest path in a graph. Here, each node in the building (such as a room, corridor, stairway entrance, etc.) is regarded as a node in the graph, and each edge on the path (such as a corridor, door, stair) has a weight, which can be adjusted according to factors such as the length of the path and the risk of fire spread. The system uses the personnel location data as the starting point, combines with the graph model of the three-dimensional evacuation path, and calculates the shortest and safest evacuation path from each employee's current location to the safety exit. This process takes into account the impact of the fire spread and avoids passing through the fire spread area. Through the Dijkstra algorithm, the optimal evacuation path for each person is dynamically calculated to ensure that the personnel at each location can reach the nearest safety exit or safe area in the shortest time. Finally, according to the calculated optimal evacuation path, evacuation guidelines are generated in real time. These guidelines can be issued to personnel through electronic display screens, broadcast systems, etc., to help them quickly find the safest evacuation route and ensure the safe evacuation of personnel.

[0045] In summary, the embodiments of the present application at least have the following technical effects:

[0046] In the embodiments of the present application, multi-source security data of campus buildings is first collected and obtained. The multi-source security data includes building structure data, fire-fighting equipment status data, environmental sensor data, and personnel distribution data. Feature extraction is performed on the multi-source security data based on preset risk dimensions to generate a fire risk feature vector. A multi-dimensional risk probability model is constructed according to historical fire case data, and risk scoring calculation is performed on the fire risk feature vector to obtain a real-time risk score. When the real-time risk score exceeds the dynamic risk threshold, a hierarchical warning instruction is triggered. An emergency treatment plan is generated according to the hierarchical warning instruction, and the fire-fighting equipment control system and the personnel evacuation guidance system are linked to perform emergency operations. These technical effects together solve the technical problems that the existing campus fire safety monitoring system lacks multi-dimensional, real-time, and comprehensive risk assessment means and is difficult to accurately judge different fire risks and respond in real time, achieving the technical effect of realizing accurate fire risk early warning and efficient emergency response through multi-source data fusion and intelligent analysis, and improving the campus fire safety management level.

[0047] Embodiment 2, based on the same inventive concept as a multi-dimensional fire safety monitoring method for a smart campus in the foregoing embodiment, as Figure 2 shown, the present application provides a multi-dimensional fire safety monitoring system for a smart campus. The system includes: a data acquisition module 11: collecting and obtaining multi-source security data of campus buildings. The multi-source security data includes building structure data, fire-fighting equipment status data, environmental sensor data, and personnel distribution data; a feature extraction module 12: performing feature extraction on the multi-source security data based on preset risk dimensions to generate a fire risk feature vector; a scoring calculation module 13: constructing a multi-dimensional risk probability model according to historical fire case data, and performing risk scoring calculation on the fire risk feature vector to obtain a real-time risk score; an instruction trigger module 14: triggering a hierarchical warning instruction when the real-time risk score exceeds the dynamic risk threshold; an emergency operation module 15: generating an emergency treatment plan according to the hierarchical warning instruction, and linking the fire-fighting equipment control system and the personnel evacuation guidance system to perform emergency operations.

[0048] Further, the feature extraction module 12 is further configured to execute the following method:

[0049] Divide the building structure data into building risk dimensions, and extract the building fire resistance rating, safety exit density, and fire compartment integrity index; divide the fire-fighting equipment status data into equipment availability dimensions, and extract equipment failure rate, fire extinguisher pressure value, and smoke detector response time; divide the environmental sensor data into environmental hazard dimensions, and extract the temperature change rate, smoke concentration gradient, and CO concentration peak; divide the personnel distribution data into personnel density dimensions, and extract the number of people on each floor, the congestion index of the escape route, and the number of people staying in key areas; perform normalization processing on the data of the building risk dimension, equipment availability dimension, environmental hazard dimension, and personnel density dimension to generate the fire risk feature vector.

[0050] Further, the scoring calculation module 13 is also used to execute the following method:

[0051] Obtain historical fire case data, and label the building structure features, equipment status features, environmental features, and personnel features corresponding to each event; use the building structure features and their building structure scores as inputs to train the building risk detection unit; use the equipment status features and their equipment status scores as inputs to train the equipment risk detection unit; use the environmental features and their environmental scores as inputs to train the environmental risk detection unit; use the personnel features and their personnel scores as inputs to train the personnel risk detection unit; integrate the building risk detection unit, equipment risk detection unit, environmental risk detection unit, and personnel risk detection unit to generate the multi-dimensional risk probability model.

[0052] Further, the scoring calculation module 13 is also used to execute the following method:

[0053] Pair the building structure features with the corresponding building structure scores to construct a training sample set; perform feature normalization and label standardization processing on the training sample set to generate standardized training data; input the standardized training data into the initial building risk detection unit and use a supervised learning algorithm for training to form a building risk detection unit capable of evaluating building structure risks.

[0054] Further, the instruction trigger module 14 is also used to execute the following method:

[0055] Obtain the campus activity scene type in the current period, and the scene types include class periods, assembly periods, and night dormancy periods; match the preset basic thresholds according to the scene types, where the basic threshold corresponds to the class period, the assembly period dynamically adjusts the basic threshold based on the personnel gathering risk, and the night dormancy period dynamically adjusts the basic threshold based on the response delay risk; superimpose the impact assessment of real-time weather conditions, and when it is detected that the meteorological environment exacerbates the fire spread risk, adaptively adjust the basic threshold to obtain the dynamic risk threshold.

[0056] Furthermore, the instruction trigger module 14 is further configured to execute the following method:

[0057] Set three levels of dynamic risk thresholds for yellow warning, orange warning, and red warning; when the real-time risk score first exceeds the yellow warning threshold, start device self-check and warning broadcast; when the score continuously exceeds the orange warning threshold for a preset duration, trigger the pre-start of the sprinkler system and unlock the access control of the escape route; when the score instantaneously exceeds the red warning threshold, perform power-off protection for the entire building and activate emergency lighting and evacuation guidance.

[0058] Furthermore, the emergency operation module 15 is further configured to execute the following method:

[0059] Obtain real-time personnel location data and fire spread data; obtain a three-dimensional evacuation path based on the building BIM model and the fire spread data; use the Dijkstra algorithm to calculate the optimal evacuation path based on the personnel location data and the three-dimensional evacuation path, and dynamically generate the evacuation guidance.

[0060] It should be noted that the above sequence of embodiments of the present application is only for description and does not represent the superiority or inferiority of the embodiments. And the above describes specific embodiments of this specification. The processes depicted in the drawings do not necessarily require the specific order and continuous order shown to achieve the desired results. In some embodiments, multitasking and parallel processing are also possible or may be advantageous.

[0061] The above are only the preferred embodiments of the present application and are not intended to limit the present application. Any modifications, equivalent replacements, improvements, etc. made within the spirit and principle of the present application shall be included in the protection scope of the present application.

[0062] This specification and the drawings are only exemplary descriptions of the present application and are considered to have covered any and all modifications, variations, combinations, or equivalents within the scope of the present application. Obviously, those skilled in the art can make various changes and modifications to the present application without departing from the scope of the present application. Thus, if these modifications and variations of the present application fall within the scope of the present application and its equivalent technologies, the present application is intended to include these changes and modifications.

Claims

1. A multi-dimensional fire safety monitoring method for smart campuses, characterized in that, The method includes: Collecting and obtaining multi-source security data of campus buildings, where the multi-source security data includes building structure data, fire equipment status data, environmental sensor data, and personnel distribution data; Extracting features from the multi-source security data based on preset risk dimensions to generate a fire risk feature vector; Constructing a multi-dimensional risk probability model according to historical fire case data, calculating a risk score for the fire risk feature vector, and obtaining a real-time risk score; When the real-time risk score exceeds the dynamic risk threshold, triggering a hierarchical warning instruction; Generating an emergency treatment plan according to the hierarchical warning instruction, and linking the fire equipment control system and the personnel evacuation guidance system to perform emergency operations.

2. The method according to claim 1, wherein The method for extracting features from the multi-source security data based on preset risk dimensions includes: Dividing the building structure data into building risk dimensions, and extracting the building fire resistance rating, safety exit density, and fire compartment integrity index; Dividing the fire equipment status data into equipment availability dimensions, and extracting the equipment failure rate, fire extinguisher pressure value, and smoke detector response time; Dividing the environmental sensor data into environmental hazard dimensions, and extracting the temperature change rate, smoke concentration gradient, and CO concentration peak; Dividing the personnel distribution data into personnel density dimensions, and extracting the number of people on each floor, the congestion index of the escape route, and the number of people staying in key areas; Normalizing the data of the building risk dimension, equipment availability dimension, environmental hazard dimension, and personnel density dimension to generate the fire risk feature vector.

3. The method according to claim 1, characterized in that, The multi-dimensional risk probability model is constructed in the following way: Obtaining historical fire case data, and labeling the building structure features, equipment status features, environmental features, and personnel features corresponding to each event; Using the building structure features and their building structure scores as inputs to train a building risk detection unit; Using the equipment status features and their equipment status scores as inputs to train an equipment risk detection unit; Using the environmental features and their environmental scores as inputs to train an environmental risk detection unit; Using the personnel features and their personnel scores as inputs to train a personnel risk detection unit; Integrating the building risk detection unit, equipment risk detection unit, environmental risk detection unit, and personnel risk detection unit to generate the multi-dimensional risk probability model.

4. The method according to claim 1, wherein The determination of the dynamic risk threshold includes: Obtaining the campus activity scene type of the current period, where the scene type includes class period, assembly period, and night dormancy period; Matching a preset basic threshold according to the scene type, where the basic threshold corresponds to the class period, the assembly period dynamically adjusts the basic threshold based on the personnel gathering risk, and the night dormancy period dynamically adjusts the basic threshold based on the response delay risk; Overlaying the impact assessment of real-time weather conditions, and adaptively adjusting the basic threshold when it is monitored that the meteorological environment exacerbates the fire spread risk to obtain the dynamic risk threshold.

5. The method according to claim 4, wherein The triggering of the hierarchical warning instruction includes: Setting three-level dynamic risk thresholds for yellow warning, orange warning, and red warning; When the real-time risk score first exceeds the yellow warning threshold, starting equipment self-check and warning broadcast; When the score continuously exceeds the orange warning threshold for a preset duration, the pre - start of the sprinkler system is triggered and the access control of the escape route is unlocked. When the score instantaneously exceeds the red warning threshold, the power protection for the whole building is executed and the emergency lighting and evacuation guidance are activated.

6. The method according to claim 5, wherein Generate an emergency treatment plan according to the classified warning instruction, including: Obtain the personnel location data and the fire spread data in real - time; Obtain the three - dimensional evacuation path based on the building BIM model and the fire spread data; Use the Dijkstra algorithm to calculate the optimal evacuation path based on the personnel location data and the three - dimensional evacuation path, and dynamically generate the evacuation guidance.

7. The method according to claim 3, wherein Train the building risk detection unit, including: Pair the building structure features with the corresponding building structure scores to construct a training sample set; Perform feature normalization and label standardization processing on the training sample set to generate standardized training data; Input the standardized training data into the initial building risk detection unit and use the supervised learning algorithm for training to form a building risk detection unit capable of evaluating the building structure risk.

8. A multi-dimensional fire safety monitoring system for smart campuses, characterized in that, The system is used to execute the multi - dimensional fire safety monitoring method for a smart campus described in any one of claims 1 - 7, including: Data acquisition module: Collect and obtain multi - source security data of campus buildings, and the multi - source security data includes building structure data, fire - fighting equipment status data, environmental sensor data, and personnel distribution data; Feature extraction module: Extract features from the multi - source security data based on preset risk dimensions to generate a fire risk feature vector; Score calculation module: Construct a multi - dimensional risk probability model according to historical fire case data, calculate the risk score for the fire risk feature vector, and obtain the real - time risk score; Instruction trigger module: When the real - time risk score exceeds the dynamic risk threshold, trigger a classified warning instruction; Emergency operation module: Generate an emergency treatment plan according to the classified warning instruction, and link the fire - fighting equipment control system and the personnel evacuation guidance system to execute emergency operations.

Images