Fire warning method and platform based on multi-data fusion

Abstract

The present application provides a multi-data fusion fire warning method and platform, which relates to the field of fire protection technology, including: collecting a multi-dimensional environmental data set of the target area; filtering the multi-dimensional environmental data set to remove noise, and obtaining a multi-dimensional environmental denoised data set; comprehensively analyzing multiple data sources based on the multi-dimensional environmental denoised data set to generate fire comprehensive data, and fusing the multi-dimensional environmental denoised data set to obtain a multi-dimensional environmental fusion data set; constructing a fire risk model based on the fire comprehensive data, synchronizing the multi-dimensional environmental fusion data set to the fire risk model according to multiple data sources for evaluation, and obtaining the fire risk level of the target area; generating a fire warning instruction based on the fire risk level, and sending it to a remote control terminal. The present application solves the technical problem that the traditional fire warning method has a single monitoring method and is easily interfered with by environmental factors, resulting in false alarms or missed alarms, and significantly improves the accuracy and response speed of fire warnings.

Description

Technical Field

[0001] The present application relates to the field of fire protection technology, and specifically to a fire warning method and platform based on multi-data fusion. Background Art

[0002] Fire warning systems play a vital role in public safety. Their primary function is to detect signs of fire early, enabling timely action and reducing fire damage. Traditional fire warning methods typically rely on single sensor devices, such as smoke or heat detectors. These devices can issue alerts upon detecting smoke or abnormal temperatures in the early stages of a fire, enabling rapid evacuation and countermeasures. However, this reliance on a single data source can lead to false or missed alerts in complex environments due to factors such as interference. For example, smoke detectors can be interfered with by kitchen smoke, while heat detectors can trigger alarms due to direct sunlight or other non-fire factors. Furthermore, traditional warning methods often rely on simple threshold assessments and lack the ability to comprehensively monitor and analyze complex environmental conditions. This makes it difficult to comprehensively assess fire risk and accurately adjust warning levels based on actual conditions. This limitation prevents firefighters from obtaining accurate and real-time environmental information when faced with fire risks, hampering their judgment and decision-making. Summary of the Invention

[0003] This application provides a multi-data fusion fire warning method and platform, which solves the technical problems of traditional fire warning methods, which have a single monitoring method and are easily interfered with by environmental factors, resulting in false alarms or missed alarms, and achieves the technical effect of significantly improving the accuracy and response speed of fire warnings.

[0004] In view of the above problems, on the one hand, the present application provides a fire warning method of multi-data fusion, which includes: collecting environmental data of a target area through multiple data sensing devices to obtain a multidimensional environmental data set; filtering the multidimensional environmental data set in combination with the fire equipment layout data of the target area to obtain a multidimensional environmental denoised data set; comprehensively analyzing multiple data sources based on the multidimensional environmental denoised data set to generate fire comprehensive data, and fusing the multidimensional environmental denoised data set according to the fire comprehensive data to obtain a multidimensional environmental fused data set; training and learning based on the fire comprehensive data to construct a fire risk model, and synchronizing the multidimensional environmental fused data set to the fire risk model for evaluation according to the multiple data sources to obtain the fire risk level of the target area; making a judgment based on the fire risk level to generate a fire warning instruction, and sending the fire warning instruction to a remote control terminal for automatic warning of the target area.

[0005] On the other hand, the present application also provides a multi-data fusion fire warning platform, which includes: an environmental data acquisition module, which is used to collect environmental data of the target area through multiple data sensing devices to obtain a multidimensional environmental data set; a data denoising module, which is used to filter out noise from the multidimensional environmental data set in combination with the fire equipment layout data of the target area to obtain a multidimensional environmental denoised data set; a data fusion module, which is used to comprehensively analyze multiple data sources based on the multidimensional environmental denoised data set to generate fire comprehensive data, and fuse the multidimensional environmental denoised data set according to the fire comprehensive data to obtain a multidimensional environmental fusion data set; a fire risk assessment module, which is used to train and learn based on the fire comprehensive data to construct a fire risk model, synchronize the multidimensional environmental fusion data set to the fire risk model for evaluation according to the multiple data sources, and obtain the fire risk level of the target area; a fire warning module, which is used to make a judgment based on the fire risk level, generate a fire warning instruction, and send the fire warning instruction to the remote control terminal for automatic warning of the target area.

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

[0007] Multiple data sensing devices are used to collect environmental data from the target area, generating a multidimensional environmental dataset. This dataset encompasses various fire risk-related information for the target area, such as temperature, smoke concentration, humidity, and other environmental parameters, providing comprehensive raw data for subsequent analysis. This multidimensional environmental dataset, combined with the firefighting equipment layout data for the target area, is then subjected to noise filtering to generate a denoised multidimensional environmental dataset. This step improves data quality and accuracy by removing irrelevant or interfering data, making subsequent analysis more reliable. Based on the denoised multidimensional environmental dataset, multiple data sources are comprehensively analyzed to generate comprehensive firefighting data. The denoised multidimensional environmental dataset is then fused based on the comprehensive firefighting data to generate a fused multidimensional environmental dataset. This step integrates information from various sources to make the data more comprehensive and representative, forming a comprehensive environmental view that more accurately reflects the fire risk situation. A fire risk model is constructed through training and learning based on the comprehensive firefighting data. The fused multidimensional environmental dataset is then synchronized with the fire risk model based on the multiple data sources for evaluation, quantifying the fire risk level in the target area and providing a key basis for subsequent early warning decisions. This step dynamically adapts to changing environmental conditions and optimizes the accuracy of fire risk predictions through machine learning. Based on the fire risk level, a fire warning instruction is generated and sent to the remote control terminal to automatically warn the target area. This step implements a rapid response mechanism, ensuring that relevant personnel receive fire risk information in a timely manner when a fire occurs and take appropriate preventive or rescue measures.

[0008] In summary, this application uses multiple data sensing devices to collect environmental data. After noise filtering, comprehensive analysis, and data fusion, a fire risk model is constructed for assessment. Finally, warning instructions are generated to achieve automated fire warning. This multi-data fusion approach overcomes the limitations of traditional single-data warnings, comprehensively considers multiple environmental factors, and eliminates data noise interference, making fire risk assessments more accurate. At the same time, automated warnings ensure timely notification of relevant personnel, effectively improving the accuracy, timeliness, and reliability of fire warnings, thereby enhancing the effectiveness of the entire fire warning system and ensuring fire safety in the target area.

[0009] The above description is only an overview of the technical solution of the present application. In order to more clearly understand the technical means of the present application, it can be implemented in accordance with the contents of the specification. In order to make the above and other purposes, features and advantages of the present application more obvious and easy to understand, the specific implementation methods of the present application are listed below. BRIEF DESCRIPTION OF THE DRAWINGS

[0010] Figure 1 A flowchart of a multi-data fusion fire warning method provided in an embodiment of the present application.

[0011] Figure 2 A schematic diagram of the process of obtaining a multi-dimensional environmental denoised dataset in the multi-data fusion fire warning method provided in an embodiment of the present application.

[0012] Figure 3 A schematic diagram of the process of generating comprehensive fire data in the multi-data fusion fire warning method provided in an embodiment of the present application.

[0013] Figure 4 A schematic diagram of the structure of the multi-data fusion fire warning platform provided in an embodiment of the present application.

[0014] Explanation of the reference numerals: environmental data acquisition module 10 , data denoising module 20 , data fusion module 30 , fire risk assessment module 40 , fire warning module 50 . DETAILED DESCRIPTION

[0015] The present application provides a multi-data fusion fire warning method and platform. This method collects environmental data through multiple sensor devices, filters out noise, conducts comprehensive analysis, and fuses the data to construct a fire risk model for assessment. Finally, it generates warning instructions to achieve automated fire warning. This application addresses the technical issues of traditional fire warning methods, which have a single monitoring method and are susceptible to interference from environmental factors, leading to false alarms or missed alarms. This method significantly improves the accuracy and response speed of fire warnings.

[0016] Example 1, as Figure 1 As shown, the embodiment of the present application provides a fire warning method based on multi-data fusion, the method comprising:

[0017] Step S1: Collect the environment of the target area through multiple data sensing devices to obtain a multi-dimensional environmental data set.

[0018] Specifically, data sensing devices refer to sensors used to collect environmental information, such as temperature sensors, humidity sensors, smoke detectors, and gas detectors. A target area refers to a specific location that needs to be monitored, such as a building, factory, or warehouse. A multidimensional environmental dataset is a collection of various environmental data collected by multiple sensors, covering multiple dimensions such as temperature, smoke concentration, and humidity.

[0019] First, a variety of data sensing devices are used to monitor the target area's environment. These sensors are placed at key locations within the target area and continuously collect environmental data. For example, in a warehouse, a temperature sensor monitors the temperature in real time, a smoke sensor detects the presence of smoke or harmful gases in the air, and a humidity sensor measures the humidity level. Each sensor transmits the collected data in a specific format (such as a digital signal) to a data processing center or storage device, ultimately forming a multidimensional environmental dataset that provides rich foundational information for subsequent analysis.

[0020] Step S2: performing noise filtering on the multidimensional environment dataset in combination with the firefighting equipment layout data of the target area to obtain a multidimensional environment denoised dataset.

[0021] Specifically, noise filtering refers to the use of specific algorithms or methods to remove irrelevant or interfering information from data to improve data quality. Firefighting equipment layout data describes the specific location distribution, type, quantity, and other related information of firefighting equipment (such as fire extinguishers and sprinkler systems). For example, this includes the specific location coordinates of fire hydrants on a specific floor, the placement of fire extinguishers, and the coverage of sprinkler systems. A multidimensional environmental denoising dataset is a data set obtained by performing noise filtering on a multidimensional environmental dataset, which can more accurately reflect the actual environmental status of the target area.

[0022] First, the firefighting equipment layout data for the target area is obtained. This data is then combined with the multidimensional environmental dataset and processed using specific algorithms, such as filtering and Kalman filtering. For example, on a certain floor, some electronic components in the fire sprinkler system may generate weak electromagnetic interference with the smoke sensor signal. Using the Kalman filtering algorithm, combined with the firefighting equipment layout data (such as the sprinkler system's location and electromagnetic characteristics), this interference data caused by the firefighting equipment is identified and removed, resulting in a denoised multidimensional environmental dataset. This noise-filtered, denoised multidimensional environmental dataset more accurately reflects the actual environmental conditions of the target area, laying the foundation for accurate fire risk assessment.

[0023] Step S3: Based on the multidimensional environment denoising dataset, a comprehensive analysis is performed on multiple data sources to generate fire protection comprehensive data, and the multidimensional environment denoising dataset is fused according to the fire protection comprehensive data to obtain a multidimensional environment fusion dataset.

[0024] Specifically, comprehensive fire protection data is generated through a comprehensive analysis of a multidimensional environmental denoised dataset and multiple data sources. This comprehensive data provides a comprehensive reflection of the safety status of the target area. The multidimensional environmental fusion dataset is derived by fusing the multidimensional environmental denoised dataset with the comprehensive fire protection data. This dataset is no longer a simple collection of sensor data; instead, through comprehensive analysis and fusion, it better reflects the relationship between the overall environmental characteristics of the target area and fire risk.

[0025] Use data analysis software or algorithm libraries to conduct in-depth mining of multiple data sources, comprehensively analyze the relationships between various environmental data, and generate comprehensive fire protection data. For example, analyze the relationship between temperature data and smoke concentration data, and consider the impact of humidity on the possibility of fire. Combine data such as temperature, humidity, and smoke concentration, and use statistical analysis methods (such as principal component analysis) to identify potential fire risk factors. Then, based on the comprehensive fire protection data, use data fusion algorithms (such as weighted average method, Bayesian estimation method, etc.) to fuse the multidimensional environmental denoising data set. Perform data fusion and integrate the scattered data into a multidimensional environmental fusion data set, making the data more comprehensive and representative, thereby more accurately reflecting the fire risk situation.

[0026] Step S4: Conduct training and learning based on the comprehensive fire protection data to construct a fire risk model. Synchronize the multi-dimensional environment fusion data set to the fire risk model according to the multiple data sources to perform evaluation and obtain the fire risk level of the target area.

[0027] Specifically, the fire risk model is constructed by analyzing historical and real-time data to assess the risk of fire in a target area. The model is trained on comprehensive firefighting data and outputs a fire risk level for the target area based on a multi-dimensional environmental fusion dataset.

[0028] Machine learning algorithms (such as decision trees and neural networks) are used to construct fire risk models using comprehensive fire data as training data. For example, when using a decision tree algorithm, various features within the comprehensive fire data (such as temperature, smoke concentration, and humidity) are used as input, and historically known fire occurrences (whether a fire has occurred, fire severity, etc.) are used as output labels for training. After the model is constructed, a multidimensional environmental fusion dataset is input into the fire risk model based on multiple data sources (i.e., previously collected and processed data sources). The algorithmic logic within the model is used for calculation and evaluation, ultimately determining the fire risk level of the target area. This level can be a numerical value (e.g., between 0 and 1, with 0 indicating no fire risk and 1 indicating an imminent fire) or a categorical level (e.g., low risk, medium risk, high risk). By inputting the multidimensional environmental fusion dataset into the model for evaluation, the fire risk level of the target area can be quantified, providing a key basis for subsequent early warning decisions.

[0029] Step S5: Based on the fire risk level, a determination is made, a fire warning instruction is generated, and the fire warning instruction is sent to the remote control terminal for automatic warning of the target area.

[0030] Specifically, fire warning instructions are response instructions generated based on the fire risk level, which can include immediately activating the sprinkler system, notifying the fire department, etc. A remote control terminal is a device or system used to receive and execute warning instructions, such as a fire monitoring center.

[0031] A judgment is made based on the fire risk level assessed in step S4 and a fire warning instruction is generated. During the judgment process, different thresholds or rules can be set to determine different operations. The generated alarm instruction is sent to the remote control terminal via a communication network (such as Ethernet, WiFi or mobile network, etc.). For example, in a large industrial park, if the fire risk level of a workshop reaches a high risk, a corresponding warning instruction is generated and sent to the remote control terminal of the monitoring center through the network of the park. The monitoring center can take corresponding fire-fighting measures according to the instruction, such as remotely starting the fire-fighting equipment in the workshop and notifying the personnel in the workshop to evacuate, etc., to achieve automated early warning of the target area.

[0032] According to the assessed fire risk level, corresponding early warning instructions are generated and automatically sent to the remote control terminal, realizing the automation of early warning operations, ensuring that relevant personnel can receive fire risk information in a timely manner, quickly take corresponding prevention or rescue measures, and effectively reduce potential losses.

[0033] Furthermore, step S1 of the embodiment of the present application further includes:

[0034] Retrieving historical environmental data archives for sensor analysis, selecting and determining multiple data sensing devices; determining multiple data sources based on the historical environmental data archives, performing collection and analysis based on the multiple data sources, and determining multiple collection points; obtaining data collection requirement information, deploying the multiple data sensing devices according to the data collection requirement information and the multiple collection points to construct a data sensing network; performing environmental traversal of the multiple data sources in the target area through the data sensing network to obtain the multidimensional environmental data set.

[0035] Specifically, the historical environmental data archive is a collection of past environmental data related to the target area, including but not limited to historical changes in temperature, humidity, smoke concentration, and other data over time. First, the historical environmental data archive for the target area is retrieved. Data analysis and data mining techniques are then used to analyze the archive to identify appropriate data sensing devices for collecting environmental data. For example, if the analysis of the historical environmental data archive reveals significant temperature fluctuations in a particular area, which correlate with fire risk, a high-precision temperature sensor may be required. If humidity is found to have a certain impact on fire risk, a humidity sensor may be selected, thereby determining multiple data sensing devices.

[0036] Next, the correlations between data in the historical environmental data archive and their relationship to fire risk are analyzed to identify multiple data sources—specific types of environmental information sources, such as temperature, humidity, smoke, and gases. For example, if historical data indicates that temperature and smoke concentration have a significant impact on fire risk, temperature sensors and smoke detectors will be prioritized. Subsequently, data collection and analysis are conducted based on the identified data sources, and specific collection points are further selected based on factors such as the layout and functional zoning of the target area. For example, within a building, sensors can be placed in kitchens, stairwells, and near electrical equipment to ensure coverage of potentially high-risk fire areas.

[0037] Obtain data collection requirements. This information can come from safety regulations for the target area, analysis of past fire incidents, or specific usage requirements, including specific requirements for collection points and sensor deployment. For example, sites housing valuable cultural relics may require extremely high accuracy and frequency in temperature and humidity measurement. Based on these data collection requirements and the multiple collection points identified, multiple data sensing devices are deployed to build a data sensor network, collecting monitoring information from multiple data sources in the target area.

[0038] Finally, through the constructed data sensing network, the target area's multiple data sources are traversed. Each data sensing device collects data from its own data source at a set collection frequency (e.g., every 5 minutes). The collected environmental data is then integrated to form the required multidimensional environmental dataset, providing an accurate and reliable data foundation for subsequent fire warnings.

[0039] Further, such as Figure 2 As shown, step S2 of the embodiment of the present application also includes:

[0040] Download a fire map of the target area, mark the target area according to the fire map, and determine the fire equipment layout data of the target area; integrate the multidimensional environmental data set with the fire equipment layout data to obtain comprehensive fire environment data; perform noise identification based on the comprehensive fire environment data to determine multiple noise sources, perform feature analysis according to the multiple noise sources, and obtain multiple noise feature data; use a filtering algorithm to combine the multiple noise feature data to filter out noise and generate initial filtered data; perform filtering verification on the initial filtered data to obtain a filtering effect, perform denoising analysis based on the filtering effect, and determine the multidimensional environmental denoising data set.

[0041] Specifically, a fire map is a map that specifically depicts fire-related information within a target area, including the location of firefighting equipment (such as fire hydrants, fire extinguishers, and sprinklers), evacuation routes, and emergency exits. First, a fire map of the target area is downloaded from the relevant system or database. The target area is then marked based on the detailed information on the fire map to determine the firefighting equipment layout data for the target area. For example, in a hotel's fire early warning system, a fire map is downloaded from the hotel's management system, and the location of fire hydrants, fire extinguishers, and other firefighting equipment layout data on each floor are marked on the map.

[0042] Use data merging tools (such as Excel or a database management system) to integrate the multidimensional environmental dataset and firefighting equipment layout data to generate comprehensive firefighting environment data. This comprehensive firefighting environment data combines the target area's environmental information and firefighting equipment layout information to more comprehensively consider various factors related to fire early warning.

[0043] Noise identification is performed based on comprehensive fire environment data. Data analysis algorithms (such as outlier detection) are combined with an understanding of the operating principles of firefighting equipment and environmental factors to identify multiple noise sources. A noise source is a source of interference that occurs during the collection of a multidimensional environmental dataset. For example, if smoke sensor data in a certain area frequently experiences abnormal fluctuations, analysis reveals that this is caused by electromagnetic radiation from a nearby fire alarm device, then the fire alarm device is considered a noise source.

[0044] Based on the identified noise sources, characteristic analysis is performed, using signal analysis tools (such as spectrum analyzer software) or statistical methods (such as calculating the mean and variance of the noise data) to obtain multiple noise characteristic data. These noise characteristic data describe the characteristics of the noise, such as its frequency range, intensity, and occurrence patterns.

[0045] A filtering algorithm is used to filter out noise by combining multiple noise signature data, separating the useful signal from the noise. This results in preliminarily cleaned data, known as the initial filtered data. Filter validation is then performed on the initial filtered data. By comparing the data before and after cleaning, the filtering effect—that is, the degree of improvement in data quality after filtering—is evaluated. The filtering effect can be assessed by comparing the data with known accurate data (if standard data samples or historically reliable data are available) or through cross-validation (dividing the data into multiple parts, one for filtering and the other for validation). For example, the filtered data can be compared with reference data collected under the same conditions but without noise interference, and the degree of reduction in data error can be calculated to evaluate the filtering effect. If the filtering effect meets the expected requirements, the filtered data can be determined to be a multidimensional environmental denoising dataset. If the filtering effect is unsatisfactory, the filtering algorithm should be adjusted or the noise signature data should be reanalyzed, and filtering should be repeated until a satisfactory multidimensional environmental denoising dataset is obtained.

[0046] The above steps integrate multidimensional environmental data and fire equipment layout data, and use statistical analysis and filtering algorithms to identify and remove noise in the data to generate a high-quality multidimensional environmental denoised dataset. This provides reliable data support for subsequent comprehensive analysis and risk assessment, thereby improving the accuracy and reliability of the fire warning system.

[0047] Furthermore, in step S2 of the embodiment of the present application, the filtering algorithm is used to combine the plurality of noise feature data to perform noise filtering to generate initial filtered data, and the method further includes:

[0048] Based on the multiple noise characteristic data, analysis is performed to obtain a fire environment noise signal; wavelet decomposition is performed on the fire environment noise signal to obtain noise signal wavelet coefficients; threshold quantization is performed based on the noise signal wavelet coefficients to determine a noise signal wavelet selection threshold; the noise signal wavelet coefficients are truncated according to the noise signal wavelet selection threshold, noise signals smaller than the noise signal wavelet selection threshold are set to zero, and effective signal information greater than the noise signal wavelet selection threshold is obtained; the effective signal information is filtered and reconstructed to obtain the initial filtered data.

[0049] Specifically, noise filtering involves first processing the raw monitoring data using data analysis tools to identify the characteristics of the interference signal and extract the noise signal from the firefighting environment. Wavelet decomposition is then performed on the firefighting environment noise signal to capture its variations over time and frequency. This decomposition is then broken down into wavelet coefficients at different scales and frequencies, representing the noise signal wavelet coefficients.

[0050] Threshold quantization is performed based on the noise signal wavelet coefficients to determine the noise signal wavelet selection threshold. This selection threshold is used to intercept and determine the noise signal wavelet coefficients in subsequent operations. In this process, statistical analysis methods can be used, such as calculating statistical quantities such as the mean and standard deviation of the noise signal wavelet coefficients, and then determining the noise signal wavelet selection threshold based on a certain rule (such as a soft threshold or hard threshold rule). For example, for a hard threshold rule, the mean plus three times the standard deviation can be used as the noise signal wavelet selection threshold.

[0051] The wavelet coefficients of the noise signal are truncated according to a selected threshold, setting noise signals below the threshold to zero while retaining valid signal information above the threshold. This process effectively removes signal components below the noise threshold, resulting in purer data. An inverse wavelet transform is performed on the filtered valid signal information, and combined with the retained wavelet coefficients, the denoised signal is restored to the time domain to obtain the initial filtered data. This data removes noise while retaining the valid signal, providing a foundation for further filtering verification and finalizing the multidimensional environmental denoising dataset.

[0052] Further, such as Figure 3 As shown, step S3 of the embodiment of the present application also includes:

[0053] Based on the multidimensional environmental denoised dataset, correlation analysis is performed on the multiple data sources to generate multiple correlation coefficients; calculations are performed according to the multiple correlation coefficients in combination with the multidimensional environmental denoised dataset to determine multiple fire indicators; based on the multiple fire indicators, the multiple data sources are evaluated to generate multiple fire source prediction scores; and based on the multiple fire source prediction scores, a comprehensive analysis is performed on the multiple data sources to generate comprehensive fire data.

[0054] Specifically, the correlation coefficient is derived from a correlation analysis of multiple data sources using a multidimensional environmental denoising dataset. It represents the degree of association between the data sources. The fire protection index is calculated using the correlation coefficient combined with the multidimensional environmental denoising dataset. It is used to measure the fire risk or status of the target area. The fire source prediction score is derived by evaluating multiple data sources using multiple fire protection indicators. It is used to predict the likelihood of a fire source. The higher the fire source prediction score for a given area, the greater the likelihood that a fire source exists in that area and may cause a fire.

[0055] Based on the multidimensional environmental denoising dataset, statistical analysis tools are used to perform correlation analysis on multiple data sources, calculating correlation coefficients between them. For example, for the temperature data source and the smoke concentration data source, the covariance between the two is calculated and then divided by their respective standard deviations to obtain the correlation coefficient representing the relationship between them. By performing this calculation between all data sources, multiple correlation coefficients are generated.

[0056] A series of firefighting indicators are determined by calculating multiple correlation coefficients combined with a multidimensional environmental denoised dataset. These indicators can be single indicators such as high temperature alarm thresholds and smoke concentration monitoring values, or comprehensive indicators such as fire hazard indicators and fire spread probability indicators. Multiple data sources are evaluated based on these multiple firefighting indicators, generating a corresponding fire source prediction score for each data source. The evaluation process utilizes pre-defined evaluation criteria (e.g., different scoring levels corresponding to the threshold range of each firefighting indicator). If a firefighting indicator represents fire hazard and exceeds a certain threshold, the data source is assigned a fire source prediction score based on pre-defined scoring rules. This evaluation of all data sources generates multiple fire source prediction scores. Additionally, machine learning models (such as support vector machines or random forests) can be used to evaluate data sources, calculating a fire source prediction score for each data source based on historical data and current monitoring values.

[0057] Finally, a comprehensive analysis is conducted across multiple data sources based on multiple fire source prediction scores, integrating and logically analyzing all fire source prediction scores and data sources. For example, higher-scoring data sources are given greater weight. By comprehensively considering the scores, data characteristics, and interrelationships of all data sources, comprehensive fire protection data is generated using rule-based expert system algorithms or decision tree algorithms from machine learning. This data fully reflects the fire situation in the target area.

[0058] Furthermore, step S3 of the embodiment of the present application further includes:

[0059] Feature extraction is performed based on the multiple fire source prediction scores to generate multiple fire comprehensive features, and a data fusion strategy is formulated based on the multiple fire comprehensive features; weighted calculation is performed on the multidimensional environment denoised dataset according to the data fusion strategy to generate multiple weight coefficients; the multidimensional environment denoised dataset is arranged in descending order based on the multiple weight coefficients to generate a denoised data sequence; the multidimensional environment denoised dataset is traversed according to the denoised data sequence to perform data fusion to obtain the multidimensional environment fused dataset.

[0060] Specifically, feature extraction is performed based on multiple fire source prediction scores. Feature extraction techniques, such as principal component analysis (PCA) or linear discriminant analysis (LDA), are used to process the fire source prediction score data, extract the features that can reflect the comprehensive fire situation, and generate multiple comprehensive fire features, such as a combination of temperature, humidity, and smoke concentration.

[0061] Next, a data fusion strategy is developed based on these comprehensive fire characteristics to fuse the multidimensional environmental denoised dataset. This data fusion strategy specifies how to weight and combine the data to achieve effective data fusion. This strategy can be adjusted based on the characteristics and importance of different data sources. For example, if historical data indicates that smoke concentration plays a significant role in fires, it can be given a higher weight during fusion.

[0062] A weighted calculation is performed on the multidimensional environmental denoising dataset according to the data fusion strategy. For example, for a multidimensional environmental denoising dataset consisting of temperature, humidity, smoke concentration, and other data, each data element is multiplied by the corresponding weight according to the weight set in the data fusion strategy (for example, the weight of smoke concentration data is 0.5, the weight of temperature data is 0.3, and the weight of humidity data is 0.2) to generate multiple weight coefficients.

[0063] Based on multiple weight coefficients, a sorting algorithm is used to sort the multidimensional environmental denoising dataset in descending order of weight coefficients to generate a denoised data sequence. The order of the data in the sequence reflects the data priority in the fusion process. The data with the highest weight coefficient is placed at the beginning of the sequence, and the data with the lowest weight coefficient is placed at the end, ensuring that the most important data is processed first.

[0064] Finally, the denoised data sequence is traversed through the multidimensional environment denoised dataset to perform data fusion, resulting in a multidimensional environment fused dataset. During this process, the data are fused sequentially from the front to the back, following the sequence order. For example, a weighted summation approach can be used: multiply the first data point in the sequence by its weight coefficient, add the second data point multiplied by its weight coefficient, and so on, ultimately resulting in a multidimensional environment fused dataset.

[0065] By extracting features and developing data fusion strategies, we achieved efficient data integration and obtained a more accurate multi-dimensional environmental fusion dataset. This process improved the comprehensiveness and practicality of the data, providing more accurate data support for fire risk assessment and early warning systems, thereby further enhancing fire safety management capabilities.

[0066] Furthermore, step S5 of the embodiment of the present application further includes:

[0067] An expected environmental threshold is set based on the fire risk level in combination with the multiple fire indicators; whether the fire risk level is greater than or equal to the expected environmental threshold is determined; if the fire risk level is greater than or equal to the expected environmental threshold, the fire risk level is extracted for risk assessment to obtain a fire risk coefficient; an impact analysis is performed based on the fire risk coefficient to generate a fire impact factor; and the fire warning instruction is generated based on the analysis of the fire impact factor in combination with the fire risk coefficient.

[0068] Specifically, the expected environmental threshold is a numerical value set based on the fire risk level and multiple fire protection indicators. It is used to determine whether the fire risk level has reached a level that requires further attention or action. The fire risk coefficient is a coefficient derived from a risk assessment of fire risk levels greater than or equal to the expected environmental threshold, more accurately quantifying the degree of fire risk. The fire impact factor is a factor derived from an impact analysis based on the fire risk coefficient. It reflects factors related to the potential impact of a fire on people, property, and the environment within the target area. Examples include the number of people potentially affected, the value of property, and the degree of pollution to the surrounding environment.

[0069] The desired environmental threshold is set based on the fire risk level and multiple fire indicators. This process requires comprehensive consideration of factors such as the type of target area (such as residential, commercial, industrial, etc.), the availability of fire protection facilities, and historical fire data.

[0070] Determine whether the fire risk level is greater than or equal to the expected environmental threshold. If so, extract the fire risk level for risk assessment. Otherwise, continue monitoring. Risk assessment can be performed using a variety of methods, such as probability-based risk assessment models or the Analytic Hierarchy Process (AHP). For example, a fire risk coefficient can be calculated based on the position of the fire risk level within a specific model and the weighting of different fire protection indicators.

[0071] Conduct an impact analysis based on the fire risk factor. This process requires incorporating specific information about the target area, such as the distribution of people and property. For example, using Geographic Information Systems (GIS) technology, we can analyze the impact of the fire risk factor on people and property within different areas, thereby generating a fire impact factor.

[0072] Fire impact factors and fire risk factors are analyzed to generate fire warning instructions. These instructions are customized based on the fire risk and impact area, such as activating sprinkler systems, sounding alarms, and directing evacuation. These instructions are then sent to a remote control terminal, providing automated warnings for the target area.

[0073] The fire risk coefficient and impact factor generated through fire risk assessment and impact analysis provide an important basis for the formulation of fire warning instructions, ensuring that fire risks can be comprehensively judged based on real-time data and historical data, and customized fire warning instructions can be generated to improve the accuracy and effectiveness of fire warnings and reduce the occurrence of potential fire accidents.

[0074] In summary, the multi-data fusion fire warning method provided in the embodiments of the present application has the following technical effects:

[0075] Multiple data sensing devices are used to collect environmental data from the target area, generating a multidimensional environmental dataset. This dataset encompasses various fire risk-related information for the target area, such as temperature, smoke concentration, humidity, and other environmental parameters, providing comprehensive raw data for subsequent analysis. This multidimensional environmental dataset, combined with the firefighting equipment layout data for the target area, is then subjected to noise filtering to generate a denoised multidimensional environmental dataset. This step improves data quality and accuracy by removing irrelevant or interfering data, making subsequent analysis more reliable. Based on the denoised multidimensional environmental dataset, multiple data sources are comprehensively analyzed to generate comprehensive firefighting data. The denoised multidimensional environmental dataset is then fused based on the comprehensive firefighting data to generate a fused multidimensional environmental dataset. This step integrates information from various sources to make the data more comprehensive and representative, forming a comprehensive environmental view that more accurately reflects the fire risk situation. A fire risk model is constructed through training and learning based on the comprehensive firefighting data. The fused multidimensional environmental dataset is then synchronized with the fire risk model based on the multiple data sources for evaluation, quantifying the fire risk level in the target area and providing a key basis for subsequent early warning decisions. This step dynamically adapts to changing environmental conditions and optimizes the accuracy of fire risk predictions through machine learning. Based on the fire risk level, a fire warning instruction is generated and sent to the remote control terminal to automatically warn the target area. This step implements a rapid response mechanism, ensuring that relevant personnel receive fire risk information in a timely manner when a fire occurs and take appropriate preventive or rescue measures.

[0076] Overall, the embodiments of this application use multiple data sensing devices to collect environmental data. After noise filtering, comprehensive analysis, and data fusion, a fire risk model is constructed for assessment. Finally, warning instructions are generated to achieve automated fire warning. This multi-data fusion approach overcomes the limitations of traditional single-data warnings, comprehensively considers multiple environmental factors, and eliminates data noise interference, making fire risk assessment more accurate. At the same time, automated warnings ensure timely notification of relevant personnel, effectively improving the accuracy, timeliness, and reliability of fire warnings, thereby enhancing the effectiveness of the entire fire warning system and ensuring fire safety in the target area.

[0077] Example 2, as Figure 4 As shown, the embodiment of the present application provides a multi-data fusion fire warning platform, which includes:

[0078] The environmental data acquisition module 10 is used to collect the environment of the target area through multiple data sensing devices to obtain a multi-dimensional environmental data set.

[0079] The data denoising module 20 is used to filter out noise from the multidimensional environment dataset combined with the firefighting equipment layout data of the target area to obtain a multidimensional environment denoised dataset.

[0080] The data fusion module 30 is used to perform a comprehensive analysis on multiple data sources based on the multidimensional environment denoising data set to generate fire comprehensive data, and fuse the multidimensional environment denoising data set according to the fire comprehensive data to obtain a multidimensional environment fusion data set.

[0081] The fire risk assessment module 40 is used to train and learn based on the comprehensive fire protection data, build a fire risk model, synchronize the multidimensional environment fusion data set to the fire risk model according to the multiple data sources, and evaluate the fire risk level of the target area.

[0082] The fire warning module 50 is used to make a judgment based on the fire risk level, generate a fire warning instruction, and send the fire warning instruction to the remote control terminal to automatically warn the target area.

[0083] Furthermore, the environmental data collection module 10 of the embodiment of the present application is further configured to perform the following steps:

[0084] Retrieving historical environmental data archives for sensor analysis, selecting and determining multiple data sensing devices; determining multiple data sources based on the historical environmental data archives, performing collection and analysis based on the multiple data sources, and determining multiple collection points; obtaining data collection requirement information, deploying the multiple data sensing devices according to the data collection requirement information and the multiple collection points to construct a data sensing network; performing environmental traversal of the multiple data sources in the target area through the data sensing network to obtain the multidimensional environmental data set.

[0085] Furthermore, the data denoising module 20 in the embodiment of the present application is further configured to perform the following steps:

[0086] Download a fire map of the target area, mark the target area according to the fire map, and determine the fire equipment layout data of the target area; integrate the multidimensional environmental data set with the fire equipment layout data to obtain comprehensive fire environment data; perform noise identification based on the comprehensive fire environment data to determine multiple noise sources, perform feature analysis according to the multiple noise sources, and obtain multiple noise feature data; use a filtering algorithm to combine the multiple noise feature data to filter out noise and generate initial filtered data; perform filtering verification on the initial filtered data to obtain a filtering effect, perform denoising analysis based on the filtering effect, and determine the multidimensional environmental denoising data set.

[0087] Furthermore, the data denoising module 20 in the embodiment of the present application is further configured to perform the following steps:

[0088] Based on the multiple noise characteristic data, analysis is performed to obtain a fire environment noise signal; wavelet decomposition is performed on the fire environment noise signal to obtain noise signal wavelet coefficients; threshold quantization is performed based on the noise signal wavelet coefficients to determine a noise signal wavelet selection threshold; the noise signal wavelet coefficients are truncated according to the noise signal wavelet selection threshold, noise signals smaller than the noise signal wavelet selection threshold are set to zero, and effective signal information greater than the noise signal wavelet selection threshold is obtained; the effective signal information is filtered and reconstructed to obtain the initial filtered data.

[0089] Furthermore, the data fusion module 30 in the embodiment of the present application is further configured to perform the following steps:

[0090] Based on the multidimensional environmental denoised dataset, correlation analysis is performed on the multiple data sources to generate multiple correlation coefficients; calculations are performed according to the multiple correlation coefficients in combination with the multidimensional environmental denoised dataset to determine multiple fire indicators; based on the multiple fire indicators, the multiple data sources are evaluated to generate multiple fire source prediction scores; and based on the multiple fire source prediction scores, a comprehensive analysis is performed on the multiple data sources to generate comprehensive fire data.

[0091] Furthermore, the data fusion module 30 in the embodiment of the present application is further configured to perform the following steps:

[0092] Feature extraction is performed based on the multiple fire source prediction scores to generate multiple fire comprehensive features, and a data fusion strategy is formulated based on the multiple fire comprehensive features; weighted calculation is performed on the multidimensional environment denoised dataset according to the data fusion strategy to generate multiple weight coefficients; the multidimensional environment denoised dataset is arranged in descending order based on the multiple weight coefficients to generate a denoised data sequence; the multidimensional environment denoised dataset is traversed according to the denoised data sequence to perform data fusion to obtain the multidimensional environment fused dataset.

[0093] Furthermore, the fire warning module 50 of the embodiment of the present application is further configured to perform the following steps:

[0094] An expected environmental threshold is set based on the fire risk level in combination with the multiple fire indicators; whether the fire risk level is greater than or equal to the expected environmental threshold is determined; if the fire risk level is greater than or equal to the expected environmental threshold, the fire risk level is extracted for risk assessment to obtain a fire risk coefficient; an impact analysis is performed based on the fire risk coefficient to generate a fire impact factor; and the fire warning instruction is generated based on the analysis of the fire impact factor in combination with the fire risk coefficient.

[0095] Through the detailed description of the multi-data fusion fire warning method in the foregoing specification, those skilled in the art can clearly understand the multi-data fusion fire warning platform in this embodiment. For the platform disclosed in Example 2, since it corresponds to the method disclosed in Example 1 and has corresponding functional modules and beneficial effects, the relevant details can be referred to the method section.

[0096] The above description of the disclosed embodiments is intended to enable one skilled in the art to implement or use the present application. Various modifications to these embodiments will be readily apparent to one skilled in the art, and the general principles defined herein may be implemented in other embodiments without departing from the spirit or scope of the present application. Therefore, the present application is not limited to the embodiments shown herein, but is intended to conform to the widest scope consistent with the principles and novel features disclosed herein.

Claims

1. A fire warning method based on multi-data fusion, characterized in that: The method comprises: The target area is collected through multiple data sensing devices to obtain a multi-dimensional environmental data set; performing noise filtering on the multidimensional environmental dataset in combination with the firefighting equipment layout data of the target area to obtain a multidimensional environmental denoised dataset; Performing a comprehensive analysis on multiple data sources based on the multidimensional environment denoised dataset to generate comprehensive fire protection data, and fusing the multidimensional environment denoised dataset based on the comprehensive fire protection data to obtain a multidimensional environment fused dataset; Conducting training and learning based on the comprehensive fire protection data to construct a fire risk model, and synchronizing the multi-dimensional environmental fusion dataset to the fire risk model for evaluation based on the multiple data sources to obtain a fire risk level for the target area; Determine based on the fire risk level, generate a fire warning instruction, and send the fire warning instruction to the remote control terminal to automatically warn the target area; Noise filtering is performed on the multidimensional environmental dataset in combination with the firefighting equipment layout data of the target area to obtain a multidimensional environmental denoised dataset, the method comprising: Downloading a fire map of the target area, marking the target area according to the fire map, and determining fire equipment layout data for the target area; The fire protection environment comprehensive data is obtained by integrating the multidimensional environment data set with the fire protection equipment layout data; performing noise identification based on the comprehensive fire environment data, determining multiple noise sources, and performing feature analysis according to the multiple noise sources to obtain multiple noise feature data; Using a filtering algorithm to combine the plurality of noise characteristic data to perform noise filtering to generate initial filtered data; Performing filtering verification on the initial filtered data to obtain a filtering effect, performing denoising analysis based on the filtering effect, and determining the multidimensional environment denoising data set; Using a filtering algorithm to combine the plurality of noise feature data to perform noise filtering to generate initial filtered data, the method includes: Analyze the noise characteristic data to obtain a fire environment noise signal; Performing wavelet decomposition on the fire environment noise signal to obtain wavelet coefficients of the noise signal; Perform threshold quantization based on the noise signal wavelet coefficient to determine the noise signal wavelet selection threshold; The noise signal wavelet coefficients are intercepted according to the noise signal wavelet selection threshold, the noise signal smaller than the noise signal wavelet selection threshold is set to zero, and the effective signal information larger than the noise signal wavelet selection threshold is obtained; Performing filtering and reconstruction on the effective signal information to obtain the initial filtered data; Comprehensively analyzing multiple data sources based on the multidimensional environmental denoising dataset to generate comprehensive fire protection data includes: performing correlation analysis on the plurality of data sources based on the multidimensional environmental denoising dataset to generate a plurality of correlation coefficients; Determine multiple fire protection indicators by calculating according to the multiple correlation coefficients in combination with the multidimensional environment denoised data set; evaluating the plurality of data sources based on the plurality of fire indicators to generate a plurality of fire source prediction scores; traversing the plurality of data sources for comprehensive analysis based on the plurality of fire source prediction scores to generate comprehensive fire protection data; The multidimensional environment denoised dataset is fused according to the fire protection comprehensive data to obtain a multidimensional environment fused dataset, the method comprising: Extracting features based on the multiple fire source prediction scores to generate multiple fire protection comprehensive features, and formulating a data fusion strategy based on the multiple fire protection comprehensive features; Performing weighted calculation on the multi-dimensional environment denoising data set according to the data fusion strategy to generate a plurality of weight coefficients; Arranging the multidimensional environment denoised data set in descending order based on the multiple weight coefficients to generate a denoised data sequence; The multidimensional environment denoised data set is traversed according to the denoised data sequence to perform data fusion to obtain the multidimensional environment fused data set.

2. The multi-data fusion fire warning method according to claim 1, characterized in that: The environment of the target area is collected by multiple data sensing devices to obtain a multi-dimensional environmental data set. The method includes: Retrieve historical environmental data archives for sensor analysis and select multiple data sensing devices; Determine multiple data sources based on the historical environmental data archive, perform collection and analysis based on the multiple data sources, and determine multiple collection points; Obtaining data collection requirement information, and deploying the plurality of data sensing devices according to the data collection requirement information and the plurality of collection points to construct a data sensing network; The multi-dimensional environment data set is obtained by performing an environmental traversal on the multiple data sources in the target area through the data sensing network.

3. The multi-data fusion fire warning method according to claim 1, characterized in that: Determining based on the fire risk level and generating a fire warning instruction, the method includes: Setting a desired environmental threshold based on the fire risk level and the plurality of fire protection indicators; Determining whether the fire risk level is greater than or equal to the expected environmental threshold; If the fire risk level is greater than or equal to the expected environmental threshold, extracting the fire risk level for risk assessment to obtain a fire risk coefficient; Performing an impact analysis based on the fire risk coefficient to generate a fire impact factor; The fire warning instruction is generated based on the analysis of the fire impact factor in combination with the fire risk coefficient.

4. The multi-data fusion fire warning platform is characterized by: The platform is used to execute the method according to any one of claims 1 to 3, comprising: An environmental data acquisition module, which is used to collect environmental data of a target area through multiple data sensing devices to obtain a multi-dimensional environmental data set; a data denoising module, configured to filter out noise from the multidimensional environmental dataset in combination with the firefighting equipment layout data of the target area to obtain a multidimensional environmental denoised dataset; a data fusion module, the data fusion module being used to perform a comprehensive analysis of multiple data sources based on the multidimensional environmental denoising dataset to generate comprehensive fire protection data, and to fuse the multidimensional environmental denoising dataset based on the comprehensive fire protection data to obtain a multidimensional environmental fused dataset; a fire risk assessment module, configured to train and learn based on the comprehensive fire protection data, construct a fire risk model, synchronize the multidimensional environmental fusion dataset with the fire risk model according to the multiple data sources, and perform an assessment to obtain a fire risk level for the target area; The fire warning module is used to make a judgment based on the fire risk level, generate a fire warning instruction, and send the fire warning instruction to the remote control terminal to automatically warn the target area.

Images