Fire risk dynamic assessment and early warning system based on K-class mean value clustering

Abstract

A fire risk dynamic assessment and early warning system based on K-class mean value clustering relates to the technical field of big data and artificial intelligence, and comprises a data acquisition module, a data processing module, a fire risk classification module, a fire risk assessment module, an early warning decision module and a feedback mechanism module. Through combination of clustering analysis and multi-factor evaluation, the system can quickly generate a risk evaluation result and timely give an early warning according to a preset threshold value. The efficient response mechanism can quickly inform related departments when the fire risk is aggravated, so that the safety of personnel and property is guaranteed.

Description

Technical Field

[0001] The present invention relates to the field of big data and artificial intelligence technology, and in particular to a fire risk dynamic assessment and early warning system based on K-mean clustering. Background Art

[0002] With the acceleration of modern urbanization, the scale and complexity of buildings and public facilities are increasing, and the importance of fire prevention and control is becoming increasingly prominent. Traditional fire warning systems usually rely on simple sensor devices such as smoke detectors and temperature sensors. These devices can send out alarm signals when a fire occurs, but due to the lack of early assessment of fire risks, effective active prevention is often not achieved. Sometimes the fire is not detected until it develops to the stage of obvious visible smoke, resulting in serious economic losses and casualties.

[0003] With the development of big data and artificial intelligence technology, data-driven fire risk assessment methods have gradually become a hot topic of research. By analyzing multi-dimensional data such as the environment, buildings, meteorological conditions, and fire-fighting equipment, the system can identify potential fire risks in advance and issue warnings when the risks accumulate to a certain level. However, the existing fire warning systems still have shortcomings in data processing. For example, some systems rely only on a single data source and fail to fully consider the complex impact of multiple risk factors, which can lead to false alarms and missed reports.

[0004] Therefore, in order to further improve the accuracy and practicality of the fire warning system, it is necessary to develop a dynamic fire risk assessment and warning method based on advanced algorithms. By introducing the K-means clustering algorithm and multi-factor model, the system can realize the automatic analysis and classification of various fire risk factors. For example, in high-risk areas, combined with data on temperature, humidity, historical fire cases and the status of firefighting facilities, the system can more accurately predict the occurrence of potential fires and issue warnings in a timely manner.

[0005] The key to fire warning systems lies in the comprehensive assessment of multiple risk factors. Traditional systems usually rely on single sensor data, such as smoke detectors and temperature sensors, which may lead to missed reports and false alarms. Therefore, modern systems need to integrate more environmental parameters, such as humidity, wind speed, temperature changes, etc., and comprehensively process and analyze multi-dimensional data to improve the accuracy and timeliness of warnings. For example, through comprehensive analysis of humidity and temperature, the system can issue early warnings to help managers take effective preventive measures.

[0006] The K-means clustering algorithm is an unsupervised learning data classification method that can classify multiple risk factors in fire risk assessment and identify areas with higher fire risks. Unlike traditional rule-based classification methods, the K-means algorithm can dynamically adjust the category center to adapt to changes in fire risks under different environmental conditions. This enables the system to identify potential hazards in a timely manner, so that preventive measures can be taken before a fire occurs.

[0007] The multi-factor model is one of the core technologies of the fire warning system. The model mathematically models multiple key factors that affect fire risk (such as temperature, humidity, historical fire records, building structure, etc.), and calculates the fire risk value through matrices and formulas. The system can adjust itself according to changes in historical data to ensure the dynamic and accurate fire risk assessment. For example, by analyzing past fire cases, the system can identify high-risk conditions and adjust the weight of risk assessment in a timely manner to enhance the system's responsiveness. Summary of the invention

[0008] In view of the above-mentioned shortcomings and deficiencies of the prior art, the present invention provides a fire risk dynamic assessment and early warning method based on K-means clustering, which aims to improve the accuracy and real-time performance of fire risk assessment in a data-driven manner, thereby effectively reducing the incidence of fire accidents.

[0009] In order to achieve the above object, the main technical solutions adopted by the present invention include:

[0010] A fire risk dynamic assessment and early warning system based on K-means clustering, including a data acquisition module, which obtains fire risk information data in real time. The data sources include building characteristics, environmental monitoring data, firefighting facility status, daily fire management records, fire drill training information, firefighting facility maintenance information, fire rescue team inspection data, and historical fire data;

[0011] The data processing module cleans, standardizes, extracts features, and preprocesses the collected data to remove noise and fill in missing values, including screening incomplete, erroneous, and duplicate data, processing outliers, standardizing data formats from different sources, and extracting features for model training;

[0012] The fire risk classification module has a K-mean clustering model to perform cluster analysis on the processed data. The model dynamically identifies and classifies the fire risks in different areas. By determining the risk cluster center, the fire risk data is divided into K categories to identify different types of risk patterns. The model iteratively updates the centroid and optimizes the distance between each data point and the nearest centroid to form a clustering result.

[0013] The fire risk assessment module has a multi-factor risk assessment model that combines building materials, climate conditions, equipment operating status, daily management, drill training records, facility maintenance, fire rescue team inspection data, and historical data, and uses a weighted algorithm to evaluate different factors to generate a comprehensive risk score;

[0014] The early warning decision module compares the generated risk score with the preset threshold to determine whether to trigger an early warning. If the risk assessment result exceeds the preset threshold, an early warning message is automatically generated and sent;

[0015] The feedback mechanism module is used to collect user feedback and system operation data to optimize and adjust model parameters.

[0016] The fire risk dynamic assessment and early warning system ensures the timeliness and accuracy of information by real-time monitoring of multiple data sources, thereby improving the effectiveness of early warning. The data processing module ensures data quality and provides a high-quality foundation for subsequent model training. Combined with the K-mean clustering model and the multi-factor risk assessment model, the system can comprehensively analyze fire risks and accurately identify different risk types. The early warning decision module automatically generates early warning information based on the set threshold to ensure timely response. At the same time, the feedback mechanism module continuously optimizes model parameters to improve the accuracy of risk assessment and response efficiency. Overall, the system effectively integrates monitoring, assessment, decision-making and feedback mechanisms to provide strong support for the identification and response of fire risks.

[0017] Furthermore, the data acquisition module is connected to sensors, monitoring equipment, and online data interfaces to acquire multi-dimensional data in real time.

[0018] Furthermore, the model in the fire risk classification module combines environmental factors, equipment status, fire management, drill training, and historical data by iteratively updating the centroid and optimizing the distance between each data point and the nearest centroid to form a clustering result.

[0019] Furthermore, in the fire risk classification module, a K-means clustering model is used to cluster the processed data, including the following steps:

[0020] Step S101, selecting a feature data set related to fire risk;

[0021] Step S102, standardizing the selected data;

[0022] Step S103: Determine the number of clusters K, and randomly select K samples as the initial centroids C k Wherein, k = 1, 2, 3…, n, n is a positive integer;

[0023] Step S104: Calculate each data point x i With all the centroids Ck The Euclidean distance of , and assign it to the nearest centroid;

[0024] Step S105, updating the position of the centroid according to the ownership of the current data point;

[0025] Step S106, judging the convergence condition, if the change of the centroid is less than the set threshold, stopping the iteration process;

[0026] Step S107: output the centroid of each cluster and the category label of each data point, and use evaluation indicators to evaluate the clustering effect.

[0027] Furthermore, the step S101 includes building characteristics, environmental monitoring data, firefighting facility status, as well as the unit's daily fire management inspection and patrol records, fire drill training conditions, firefighting facility maintenance information, fire rescue team inspection data, and temperature and humidity variables.

[0028] Furthermore, in step S102, the standardization process includes Z-score standardization and Min-Max scaling.

[0029] Furthermore, in step S103, the number of clusters K is determined by selecting the elbow method or the silhouette coefficient method.

[0030] To ensure the accuracy of the early warning system, the system needs to continuously monitor new data and regularly update the clustering model. This dynamic update capability enables the early warning system to better adapt to environmental changes and the impact of new data.

[0031] Through the above steps, the K-means algorithm provides an effective method for fire risk assessment, which can timely identify potential fire risks and issue early warnings, thereby enhancing the ability of fire prevention and control. The K-means clustering model has significant advantages in fire risk assessment. First, it can effectively identify and classify different types of fire risk patterns, thereby providing a quantitative basis for the early warning system. By clustering fire risk data, the risk distribution and characteristics can be understood more clearly, enabling decision makers to formulate targeted response measures. Secondly, the model shows high efficiency in processing large-scale data, and can quickly iterate and update the centroid, thereby improving the accuracy of real-time risk assessment. Finally, the results of K-means clustering can be combined with other evaluation methods to provide a reliable foundation for the multi-factor risk assessment model and enhance the overall reliability and responsiveness of the system.

[0032] Furthermore, in the fire risk assessment module, a multi-factor risk assessment model is used to evaluate different factors using a weighted algorithm to generate a comprehensive risk score, including the following steps:

[0033] Step S201, selecting factors related to fire risk;

[0034] Step S202: Clean and preprocess the collected data, including removing missing values ​​and outliers, and standardizing data from different sources;

[0035] Step S203: using a correlation analysis method to screen out key features that are closely related to fire risk;

[0036] Step S204, using multi-factor analysis to reduce dimensionality;

[0037] Step S205: selecting an evaluation model, including a multiple linear regression model;

[0038] Step S206: Use the training data set to train the model, adjust its parameters, input real-time data into the model for risk assessment, and calculate the fire risk score;

[0039] Step S207: set a risk threshold, output the risk assessment result, and determine the warning level based on the calculated score.

[0040] Furthermore, in step S207, the warning level includes a fire risk score of R < 0.3 for low risk, 0.3 ≤ R < 0.7 for medium risk, and 0.7 ≤ R < 1.0 for high risk.

[0041] This multi-factor risk assessment method can comprehensively consider various factors that affect fire risk and ensure the accuracy and reliability of the assessment results. Data cleaning and preprocessing steps improve data quality, making subsequent analysis more effective. The use of correlation analysis and dimensionality reduction techniques helps to extract the most important features, thereby simplifying the model and improving prediction performance. This method can dynamically adjust model parameters, adapt to real-time data input, achieve accurate fire risk score assessment, and ensure timely and effective early warning, greatly improving the efficiency and safety of fire prevention and control.

[0042] The fire risk warning system of the present invention can not only realize real-time and accurate fire risk assessment, but also effectively reduce false alarms and missed alarms, improve fire response capabilities, and provide strong technical support for preventing fire risks.

[0043] The beneficial effects of the present invention are:

[0044] In the fire risk dynamic assessment and early warning system of the present invention, integrating multiple data sources can effectively improve the system's monitoring coverage and fire prediction accuracy.

[0045] The present invention uses the K-means clustering model to effectively identify potential fire risk areas. The model clusters fire risk factors with similar characteristics based on historical data and real-time monitoring data, thereby helping managers quickly locate risk points and take preventive measures in advance.

[0046] The multi-factor risk assessment model can comprehensively consider multiple influencing factors and quantitatively analyze the contribution of each factor to fire risk. This method makes risk assessment more comprehensive and objective, improves the accuracy of the assessment, and ensures that risk judgments in different scenarios can be effectively supported.

[0047] By combining cluster analysis and multi-factor assessment, the system can quickly generate risk assessment results and issue early warnings in a timely manner according to preset thresholds. This efficient response mechanism can quickly notify relevant departments when the fire risk increases, ensuring the safety of personnel and property.

[0048] Through the systematic risk assessment and early warning mechanism, the present invention enables decision makers to formulate scientific emergency response strategies based on the information output by the model, which not only improves management efficiency, but also optimizes the allocation of resources, making the measures to deal with fire more effective and reasonable. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 It is a schematic diagram of the structure of the fire risk dynamic assessment and early warning system based on K-means clustering of the present invention;

[0050] Figure 2 It is a flow chart of the fire monitoring and early warning process of the present invention regarding the K-type mean algorithm;

[0051] Figure 3 It is a flow chart of the fire monitoring and early warning method of the present invention regarding the multi-factor model. DETAILED DESCRIPTION

[0052] In order to better explain the present invention and facilitate understanding, the present invention is described in detail below through specific implementation modes in conjunction with the accompanying drawings.

[0053] The present invention provides a fire risk dynamic assessment and early warning system based on K-means clustering, such as Figure 1 As shown, it includes a data acquisition module, a data processing module, a fire risk classification module, a fire risk assessment module, a warning decision module, and a feedback mechanism module.

[0054] The data acquisition module acquires fire risk information data in real time, and the data sources include building characteristics, environmental monitoring data, firefighting facility status, daily fire management records (patrol inspection records), fire drill training information, firefighting facility maintenance information, fire rescue team inspection data, and historical fire data. The data acquisition module is connected to sensors, monitoring equipment, and online data interfaces to acquire multi-dimensional data in real time to ensure the timeliness and accuracy of information.

[0055] Sensors are used to monitor environmental changes and can collect fire-related data such as temperature, humidity, and smoke concentration in real time. Monitoring equipment includes cameras, smoke detectors, etc., which can provide visual and perceptual information to help identify potential fire risks. Online data interfaces are interfaces that connect to other systems or databases to obtain historical data and real-time information. The data of daily fire management and patrol inspections of the unit reflects the daily safety status, fire drill training records provide an assessment of emergency response capabilities, and fire facility maintenance information ensures the integrity and response effectiveness of the facilities.

[0056] The data processing module cleans, standardizes, extracts features, and preprocesses the collected data to remove noise and fill missing values, thereby generating high-quality feature vectors. This process includes screening incomplete, erroneous, and duplicate data, processing outliers, standardizing data formats from different sources, and extracting features for model training. This process provides a basis for subsequent model training. For example, the analysis of historical fire cases will help identify key risk factors and further improve the accuracy of the model.

[0057] Data cleaning includes screening the collected raw data and removing incomplete, erroneous or duplicate data to improve data quality. Standardization is the conversion of data from different sources into a unified standard format to ensure comparability between different data sets. Feature extraction is the identification and extraction of features useful for model training from raw data, such as the time, location, environmental conditions, equipment status, daily management, training and drill frequency, problems found in external supervision and inspection, and historical fire risks.

[0058] The fire risk classification module has a K-mean clustering model, which performs cluster analysis on the processed data. The model dynamically identifies and classifies fire risks in different areas, and monitors and evaluates high-risk areas by determining the risk clustering center. By incorporating data from historical fire cases into the model, the accuracy of clustering can be improved, making the system more targeted. The fire risk data is divided into K categories to identify different types of risk patterns. The model iteratively updates the centroid and optimizes the distance between each data point and the nearest centroid to form a clustering result. This process not only takes into account environmental factors and equipment status, but also combines the unit's fire management, drill training and historical cases to provide a quantitative basis for risk warning.

[0059] Specifically, in the fire risk classification module, the K-means clustering model is used to cluster the processed data, such as Figure 2 As shown, the following steps are included:

[0060] Step S101, select a feature data set related to fire risk; including building characteristics, environmental monitoring data, firefighting facility status, as well as the unit's daily fire management inspection and patrol records, fire drill training, firefighting facility maintenance information, fire rescue team inspection data, as well as temperature, humidity, smoke concentration, building structure characteristics, etc.

[0061] Step S102: Standardize the selected data to eliminate the influence of different dimensions on the clustering results. Common standardization methods include:

[0062] Z-score normalization:

[0063] Where: μ represents the sample mean, σ represents the sample variance.

[0064] Min-Max Scaling:

[0065] Step S103: Determine the number of clusters K, which can be selected by the elbow method or the silhouette coefficient method, and randomly select K samples as the initial centroid C k Wherein, k = 1, 2, 3…, n, n is a positive integer;

[0066] Step S104: Calculate each data point x i With all the centroids C k The Euclidean distance of , and assign it to the nearest centroid;

[0067]

[0068] Among them, the Euclidean distance ||x i -C k ||The calculation formula is:

[0069]

[0070] Where: m represents the number of features.

[0071] Step S105, updating the position of the centroid according to the ownership of the current data point;

[0072]

[0073] Among them, N k is the number of data points assigned to cluster k, S k is the set of data points in cluster k.

[0074] Step S106, judging the convergence condition, if the change of the centroid is less than the set threshold ε, then stopping the iteration process;

[0075]

[0076] Step S107: Output the centroid C of each cluster k And the category label of each data point, and use evaluation indicators (such as silhouette coefficient) to evaluate the clustering effect.

[0077] K-means clustering is an unsupervised learning algorithm that divides data into K classes, optimizes similarities within classes and differences between classes, and identifies patterns and structures in data. The centroid is the center point of each cluster, representing the average value of all data points in that class. The algorithm iteratively updates the centroid position to optimize the clustering results.

[0078] The fire risk assessment module has a multi-factor risk assessment model, which combines building materials, climate conditions, equipment operating status, daily management, drill training records, facility maintenance, fire rescue team inspection data, and historical data, and uses a weighted algorithm to evaluate different factors to generate a comprehensive risk score. The multi-factor risk assessment model can accurately reflect the complexity of fire risks and support scientific decision-making.

[0079] The multi-factor model takes into account multiple factors and evaluates the overall risk through weighted analysis. This method can more comprehensively reflect complex risk situations. The weighted algorithm assigns different weights to each factor based on its importance to calculate the comprehensive risk score.

[0080] Specifically, in the fire risk assessment module, a multi-factor risk assessment model is used to evaluate different factors using a weighted algorithm to generate a comprehensive risk score, such as Figure 3 As shown, the following steps are included:

[0081] Step S201: Select factors related to fire risk; including but not limited to:

[0082] Environmental factors, such as temperature T, humidity H, and wind speed W.

[0083] Building structural characteristics, including building materials M, number of floors L, and purpose U.

[0084] Fire history data, such as historical fire frequency F and severity S.

[0085] The status of fire-fighting facilities, including the number of fire extinguishers E and the unobstructed condition of fire-fighting passages C.

[0086] Step S202: Clean and preprocess the collected data, including removing missing values ​​and outliers, and standardizing data from different sources to ensure data quality and consistency.

[0087] Step S203: Use a correlation analysis method to screen out key features that are closely related to fire risk. Specifically, use a Pearson correlation coefficient analysis method to select key features that are related to fire risk.

[0088] Step S204: Dimensionality reduction is performed using multi-factor analysis to extract the main influencing factors.

[0089] Step S205: Select an evaluation model, including a multiple linear regression model.

[0090] The formula used in the multiple linear regression model is as follows:

[0091] R=β 0 +β 1 T+β 2 H+β 3 W+β 4 M+β 5 L+β 6 U+β 7 F+β 8 S+β 9 E+β 10 C + ε (Formula 7);

[0092] Among them, R is the fire risk score, β i is the regression coefficient and ε is the error term.

[0093] Step S206: Use the training data set to train the model and adjust its parameters to improve prediction accuracy. Input the real-time data into the model for risk assessment and calculate the fire risk score.

[0094] R=f(T,H,W,M,L,U,F,S,E,C) (Formula 8);

[0095] Among them, f is the trained model function, which represents the relationship between various factors and fire risk.

[0096] Step S207: set a risk threshold, output the risk assessment result, and determine the warning level based on the calculated score.

[0097]

[0098] The warning decision module compares the generated risk score with a preset threshold to determine whether to trigger a warning. If the risk assessment result exceeds the preset threshold, a warning message is automatically generated and sent.

[0099] After the risk assessment, the early warning decision module will make a judgment based on the set threshold to decide whether to trigger the early warning. Once the risk assessment result exceeds the threshold, the module will automatically generate early warning information and quickly notify relevant personnel through SMS, email, etc. to ensure timely response.

[0100] The warning threshold is a risk assessment standard set based on historical data and actual conditions. Once the assessment result exceeds this value, the system will trigger a warning. The warning information includes notifications of risk assessment results and recommended response measures, which are used to inform relevant personnel in a timely manner.

[0101] The feedback mechanism module is used to collect user feedback and system operation data to optimize and adjust model parameters, helping the system to continuously optimize and adjust model parameters, improve the accuracy of risk assessment and the response efficiency of early warning. Through the feedback mechanism, the system can continuously adapt to changes in actual applications and enhance reliability.

[0102] The feedback mechanism is used to collect operational feedback from users and the system in order to continuously optimize system performance based on actual conditions. Model parameters are key variables that affect the model's predictive ability, and adjusting these parameters can improve the model's accuracy and robustness.

[0103] Through the collaborative work of the above modules, the early warning system can achieve efficient monitoring and early warning of fire risks, providing strong support for fire prevention. The perfect application of these modules, especially the introduction of multi-factor models and clustering methods, enables the system to comprehensively, accurately and dynamically reflect the fire risk status of the unit, and make real-time adjustments based on actual management conditions and environmental changes to achieve more efficient fire risk early warning.

[0104] In the fire warning system of the present invention, the integration of multiple data sources can effectively improve the monitoring coverage and fire prediction accuracy of the system. First, the daily fire management data of the unit provides the content and results of daily inspections. By identifying potential hidden dangers, the system can grasp the fire prevention status of the unit in real time.

[0105] Information such as the frequency and participation of fire drills and training can help assess the emergency preparedness of a unit. Frequent fire drills and training often mean stronger emergency response capabilities. Combining these data can provide a reference for risk assessment at the unit behavior management level.

[0106] The maintenance data of fire protection facilities is equally important, as it ensures that the facilities are always in good condition. If a facility has repair or maintenance defects, the system will increase the fire risk score of the area where the facility is located, thereby improving the effectiveness of early warning.

[0107] The system collects fire supervision and inspection data and records any problems found by fire rescue teams during inspections of units. This type of data can effectively update the system risk score, further strengthen the model's real-time control of fire hazards, and ensure continuous improvement of the unit's fire safety.

[0108] The analysis of historical fire cases provides the model with rich empirical data. The system can extract key features and risk factors from similar cases to optimize risk identification and analysis capabilities. The addition of these data sources empowers the fire warning system, enabling it to dynamically and comprehensively assess unit fire risks and improve the accuracy and response speed of warnings.

[0109] Through the K-means clustering model, the system can effectively identify potential fire risk areas. Based on historical data and real-time monitoring data, the model clusters fire risk factors with similar characteristics together, helping managers quickly locate risk points and take preventive measures in advance.

[0110] The multi-factor risk assessment model can comprehensively consider multiple influencing factors and quantitatively analyze the contribution of each factor to fire risk. This method makes risk assessment more comprehensive and objective, improves the accuracy of the assessment, and ensures that risk judgments in different scenarios can be effectively supported.

[0111] Both models can be dynamically adjusted and optimized based on new data to ensure the timeliness and adaptability of the models in practical applications. As new data is continuously introduced, the models can quickly learn and adapt to new risk characteristics, thereby maintaining a high level of assessment accuracy.

[0112] By combining cluster analysis and multi-factor assessment, the system can quickly generate risk assessment results and issue warnings in a timely manner based on preset thresholds. This efficient response mechanism can quickly notify relevant departments when the fire risk increases, ensuring the safety of people and property.

[0113] Through systematic risk assessment and early warning mechanisms, decision makers can formulate scientific emergency response strategies based on the information output by the model. This not only improves management efficiency, but also optimizes resource allocation, making fire response measures more effective and reasonable.

[0114] Although the embodiments of the present invention have been shown and described above, it is to be understood that the above embodiments are illustrative and are not to be construed as limitations on the present invention. Alterations, modifications, substitutions and variations of the above embodiments by a person skilled in the art are all within the scope of the present invention.

Claims

1. A fire risk dynamic assessment and early warning system based on K-means clustering, characterized by: It includes a data acquisition module to obtain fire risk information data in real time. The data sources include building characteristics, environmental monitoring data, firefighting facility status, daily fire management records, fire drill training information, firefighting facility maintenance information, fire rescue team inspection data, and historical fire data; The data processing module cleans, standardizes, extracts features, and preprocesses the collected data to remove noise and fill in missing values, including screening incomplete, erroneous, and duplicate data, processing outliers, standardizing data formats from different sources, and extracting features for model training; The fire risk classification module has a K-mean clustering model to perform cluster analysis on the processed data. The model dynamically identifies and classifies the fire risks in different areas. By determining the risk cluster center, the fire risk data is divided into K categories to identify different types of risk patterns. The model iteratively updates the centroid and optimizes the distance between each data point and the nearest centroid to form a clustering result. The fire risk assessment module has a multi-factor risk assessment model that combines building materials, climate conditions, equipment operating status, daily management, drill training records, facility maintenance, fire rescue team inspection data, and historical data, and uses a weighted algorithm to evaluate different factors to generate a comprehensive risk score; The early warning decision module compares the generated risk score with the preset threshold to determine whether to trigger an early warning. If the risk assessment result exceeds the preset threshold, an early warning message is automatically generated and sent; The feedback mechanism module is used to collect user feedback and system operation data to optimize and adjust model parameters.

2. A fire risk dynamic assessment and early warning system based on K-means clustering as claimed in claim 1, characterized in that: The data acquisition module is connected to sensors, monitoring equipment, and online data interfaces to acquire multi-dimensional data in real time.

3. A fire risk dynamic assessment and early warning system based on K-means clustering as claimed in claim 1, characterized in that: The model in the fire risk classification module forms a clustering result by iteratively updating the centroid and optimizing the distance between each data point and the nearest centroid, combining environmental factors, equipment status, fire management, drill training, and historical data.

4. A fire risk dynamic assessment and early warning system based on K-means clustering as claimed in claim 1, characterized in that: In the fire risk classification module, the K-means clustering model is used to cluster the processed data, including the following steps: Step S101, selecting a feature data set related to fire risk; Step S102, standardizing the selected data; Step S103: Determine the number of clusters K, and randomly select K samples as the initial centroids C k Wherein, k = 1, 2, 3…, n, n is a positive integer; Step S104: Calculate each data point x i With all the centroids C k The Euclidean distance of , and assign it to the nearest centroid; Step S105, updating the position of the centroid according to the ownership of the current data point; Step S106, judging the convergence condition, if the change of the centroid is less than the set threshold, stopping the iteration process; Step S107: output the centroid of each cluster and the category label of each data point, and use evaluation indicators to evaluate the clustering effect.

5. A fire risk dynamic assessment and early warning system based on K-means clustering as claimed in claim 4, characterized in that: The step S101 includes building characteristics, environmental monitoring data, firefighting facility status, the unit's daily fire management inspection and patrol records, fire drill training, firefighting facility maintenance information, fire rescue team inspection data, and temperature and humidity variables.

6. A fire risk dynamic assessment and early warning system based on K-means clustering as claimed in claim 4, characterized in that: In step S102, the standardization process includes Z-score standardization and Min-Max scaling.

7. The fire risk dynamic assessment and early warning system based on K-means clustering according to claim 1, characterized in that: In step S103, the number of clusters K is determined by selecting the elbow method or the silhouette coefficient method.

8. The fire risk dynamic assessment and early warning system based on K-means clustering according to claim 1, characterized in that: In the fire risk assessment module, a multi-factor risk assessment model is used to evaluate different factors using a weighted algorithm to generate a comprehensive risk score, including the following steps: Step S201, selecting factors related to fire risk; Step S202: Clean and preprocess the collected data, including removing missing values ​​and outliers, and standardizing data from different sources; Step S203: using a correlation analysis method to screen out key features that are closely related to fire risk; Step S204, using multi-factor analysis to reduce dimensionality; Step S205: selecting an evaluation model, including a multiple linear regression model; Step S206: Use the training data set to train the model, adjust its parameters, input real-time data into the model for risk assessment, and calculate the fire risk score; Step S207: set a risk threshold, output the risk assessment result, and determine the warning level based on the calculated score.

9. A fire risk dynamic assessment and early warning system based on K-means clustering as claimed in claim 8, characterized in that: In step S207, the warning level includes a fire risk score of R < 0.3 for low risk, 0.3 ≤ R < 0.7 for medium risk, and 0.7 ≤ R < 1.0 for high risk.

Images