Fire monitoring and identification system based on multi-sensor fusion

Abstract

The invention relates to the technical field of data analysis, in particular to a fire monitoring and recognition system based on multi-sensor fusion. Comprising a multi-modal sensor data fusion processing module, a fire occurrence probability intelligent evaluation module, a fire occurrence probability real-time prediction module, a multi-level detection and feedback module and an automatic control and emergency response module. Extracting multi-modal fire data fusion features through a data fusion algorithm to capture fire signs; a three-dimensional fire monitoring model is created, and model parameters and performance evaluation are adjusted, so that the fire risk is quickly judged; the probability of fire occurrence in the current monitoring area state is predicted, so that emergency measures can be planned in advance subsequently; fire identification is carried out through a multi-level verification mechanism, the reliability of the system is improved, and the response speed is improved by dividing fire grades and constructing an emergency response strategy library. The invention is used for solving the technical problems of incomplete monitoring and insufficient management and control in current fire prevention and management.

Description

Technical Field

[0001] The present invention relates to the technical field of data analysis, and in particular to a fire monitoring and identification system based on multi-sensor fusion. Background Art

[0002] Data analytics is a broad and deep field that involves the collection, processing, modeling, interpretation and presentation of data to extract meaningful information and help understand complex data sets. This information can then be used to make informed decisions, predict trends, identify patterns, optimize processes or discover new insights.

[0003] With the continuous advancement of technology and the continuous expansion of application fields, the public safety field is also developing steadily. However, in fire prevention and control, the following problems still exist: single monitoring means, traditional fire monitoring systems usually rely on a single or a few sensors, which are easily affected by environmental interference, sensor failure or diversity of fire characteristics, and may not be able to fully capture the various precursor information of fire, resulting in insufficient monitoring accuracy and reliability; fire risk assessment and identification are not accurate, traditional fire risk assessment often relies on manual experience and post-analysis, and cannot reflect the changes in fire risk in real time and dynamically, thus affecting the timeliness and accuracy of fire warning, and the fire identification process may be interfered by various factors, such as false alarms, missed alarms, etc.; emergency response lags: after a fire occurs, traditional emergency response mechanisms often rely on manual judgment and decision-making, resulting in a long response time, unable to quickly and effectively control the spread of the fire, and increasing the risk of casualties and property losses. To this end, the present invention proposes a fire monitoring and identification system based on multi-sensor fusion. Summary of the invention

[0004] The purpose of the present invention is to solve the problems in the background technology and to propose a fire monitoring and identification system based on multi-sensor fusion.

[0005] In order to achieve the above object, the present invention adopts the following technical solutions: The fire monitoring and identification system based on multi-sensor fusion includes: multi-modal sensor data fusion processing module, fire possibility intelligent assessment module, fire probability real-time prediction module, multi-level detection and feedback module, and automatic control and emergency response module; The multimodal sensor data fusion processing module is used to collect raw data by integrating multiple sensors, and extract multimodal fire data fusion features in combination with data fusion processing algorithms; wherein the multiple sensors include but are not limited to temperature sensors, smoke sensors, flame detectors, infrared thermal imagers, etc.; The fire occurrence possibility intelligent assessment module is used to create a three-dimensional fire monitoring model based on the extracted multi-modal fire data fusion features, and to adjust parameters and evaluate the performance of the created three-dimensional fire monitoring model; The fire occurrence probability real-time prediction module is used to introduce real-time monitoring area parameters as input for system judgment, and predict the probability of fire occurrence in the current monitoring area state through the fire occurrence possibility assessment module; The multi-level detection and feedback module is used to set up a multi-level verification mechanism, identify fires through the multi-level verification mechanism, and provide feedback on fire identification. The automatic control and emergency response module is used to classify fire hazard levels and build an emergency response strategy library based on fire identification conditions, and then automatically select the optimal emergency response strategy.

[0006] It should be noted that the application object of the fire monitoring and identification system based on multi-sensor fusion in the embodiment of the present invention can be fire prevention and control in the field of public safety. Specifically, it can be to comprehensively analyze and process the collected multi-source sensor data through artificial intelligence and data analysis technology to achieve real-time monitoring and timely regulation of fire risks. Through data preprocessing and fusion analysis of the multimodal sensor data fusion processing module, model creation and evaluation of the fire possibility intelligent evaluation module, accurate prediction of the fire probability real-time prediction module, effective verification of the multi-level detection and feedback module, and optimal strategy selection of the automatic control and emergency response module, a highly integrated and intelligent fire monitoring and identification system is formed, so that the system can warn of fire risks in advance, respond to fire incidents in time, and ensure the safety of life and property of personnel; at the same time, the system also promotes the intelligent development of fire protection technology, meets the needs of efficient and accurate fire prevention and control, and realizes the transformation from passive response to active prevention. It is of great significance to improve the level of social public safety, optimize the allocation of emergency resources, and promote the construction of smart cities.

[0007] Furthermore, the multimodal sensor data fusion processing module collects raw data by integrating multiple sensors and extracts multimodal fire data fusion features by combining the data fusion processing algorithm, including: Select the corresponding sensor type and location based on the characteristics of the monitoring area, including the size of the space, the type of potential fire source, and the complexity of the environment. For example, increase the density of smoke sensors in areas with dense flammable materials, and deploy infrared thermal imagers in open areas to monitor distant fire sources. Raw data is periodically obtained from each sensor and labeled ,in, Represents the sensor index variable, used to traverse the sensor; Define the sensor set as ,in, Indicates the total number of sensors; For each sensor , whose original data is represented as a time series ,in, Indicates time, Indicates sensor In time The total number of data points collected; For each data point , set the data cleaning threshold To clean: In the formula, Represents the data point index variable, used to traverse the data points; Respectively represent sensors The mean and standard deviation of the data; Indicates invalid data; Integrate the preprocessed multi-source sensor data, perform fusion processing on the acquired multi-source sensor data, and extract the multi-modal fire data fusion features; at the same time, according to the accuracy and reliability of each sensor, dynamically allocate its weight in the fusion process; the specific expression is: In the formula, Represents the multimodal fire data fusion feature, that is, a comprehensive feature vector containing information from multiple sensors, which is used to reflect the various characteristics of the fire, such as temperature, smoke concentration, the presence or absence of flames, etc. Represents the sensor index variable; Indicates the total number of sensors; It's time The weight of the dependency, indicating the sensor In time accuracy and reliability; Indicates that the slave sensor Functions that extract features from data; Respectively indicate adjustment of the sensor Coefficient of accuracy and reliability; Respectively represent sensors At current time The accuracy evaluation value and reliability evaluation value of

[0008] Furthermore, the fire occurrence possibility intelligent assessment module creates a three-dimensional fire monitoring model based on the extracted multi-modal fire data fusion features, and the process of parameter adjustment and performance evaluation of the created three-dimensional fire monitoring model includes: Get the multimodal fire data fusion feature set: in, Represents the acquired multi-modal fire data fusion feature index; Indicates at time The multi-modal fire data fusion feature values ​​obtained above; Represents the number of multimodal fire data fusion features; Define the set of environmental characteristics of the monitoring area: in, Indicates the number of environmental features in the monitored area; Building a 3D fire monitoring model , where Represents the parameters of the three-dimensional fire monitoring model; Through the loss function Measures the difference between the model prediction and the actual fire occurrence. Fire monitoring is a binary classification problem, that is, whether the fire occurs or not. The specific expression is; In the formula, Represents the total number of samples selected from the historical fire data; Indicates the fire status label, used to describe the time The fire status, where 0 means no fire and 1 means fire; represents the probability of fire occurrence predicted by the three-dimensional fire monitoring model, where ; Updating parameters of 3D fire monitoring model using gradient descent method , to minimize the loss function , where the specific expression is: In the formula, represents the updated three-dimensional fire monitoring model parameters; represents an iteration index, i.e., describes the number of iterations or stages in the training process of the three-dimensional fire monitoring model, and is used to iteratively update the parameters of the three-dimensional fire monitoring model during the training process; Indicates Parameters of the three-dimensional fire monitoring model for the first iteration; Represents the learning rate, which is used to control the step size of parameter adjustment; It means that the loss function is The gradient at The cross-validation method was used to evaluate the performance of the three-dimensional fire monitoring model.

[0009] Furthermore, the process of the fire occurrence possibility intelligent assessment module using a cross-validation method to evaluate the performance of the three-dimensional fire monitoring model includes: A1. Extract a complete dataset containing multimodal fire data fusion features and fire status labels from historical data , where the dataset Each record in Including multi-modal fire data fusion feature set and the corresponding fire status label ; Understandably, the complete dataset It collects the multimodal fire data fusion feature set and the corresponding fire status label, that is, each piece of data not only contains the multimodal fire data fusion feature of the monitoring area, but also contains the label information of whether a fire has occurred, and 0 means no fire and 1 means fire; A2. Dataset Randomly divide into Mutually exclusive subsets ;in, The value is 5 or 10 and is called Fold cross validation; each subset contains the same amount of data, and each subset should maintain the representativeness of the data, that is, contain samples with and without fire conditions; A3. For each subset , which is used as the validation set, and the rest The subsets are used as training sets to train the three-dimensional fire monitoring model; among them, represents the index of the subset, and ; A4. During the training of the 3D fire monitoring model, the training set data is used to learn the model parameters, and the validation set data is used to learn the model parameters. Test the performance of the model on A5. Repeat steps A3-A4 until each subset is used as a validation set. A6. After each verification, a set of performance indicators is obtained. The performance indicators obtained from the verification are averaged to obtain the performance of the three-dimensional fire monitoring model in the data set. Comprehensive performance evaluation on .

[0010] Furthermore, the fire occurrence probability real-time prediction module introduces real-time monitoring area parameters as input for system determination. The process of predicting the probability of fire occurrence in the current monitoring area state through the fire occurrence possibility assessment module includes: make For at the moment The real-time monitoring area index vector includes the current multi-modal fire data fusion characteristics and the environmental characteristics of the current monitoring area; Real-time monitoring of regional indicator vectors Input the fire probability assessment module to predict the probability of fire in the current monitoring area, and then obtain the real-time prediction result of fire probability. .

[0011] Furthermore, the multi-level detection and feedback module sets a multi-level verification mechanism, and the process of performing fire identification through the multi-level verification mechanism and feeding back the fire identification situation includes: Set up a multi-level verification mechanism, including primary testing, intermediate analysis and advanced confirmation; The process of fire identification through a multi-level verification mechanism is as follows: Preset primary detection threshold , if the fire probability is predicted in real time The value exceeds the primary detection threshold , then enter the intermediate analysis; otherwise, identify that there is no fire risk in the current monitoring area; Real-time prediction of results by calculating the probability of fire Rate of change : In the formula, Indicates at time The real-time prediction result value of the probability of fire occurrence; Preset intermediate detection threshold ,like Continues to increase and exceeds the intermediate detection threshold , then enter advanced confirmation; otherwise, identify the current monitoring area as a possible fire risk, requiring further analysis by maintenance personnel; Assume that the Moment Real-time prediction results of multiple fire occurrence probabilities and calculation of their cumulative effect statistics : In the formula, Indicates Moment The cumulative effect statistics of the real-time prediction results of multiple fire occurrence probabilities are used to monitor the cumulative changes in the process; Indicates Moment The cumulative effect statistics of the real-time prediction results of multiple fire occurrence probabilities provide the historical background of the cumulative effect, making the current moment The calculation of the cumulative effect takes into account all previous changes; Indicates at time Real-time prediction result value of multiple fire occurrence probability; It represents the benchmark value of the real-time prediction of the probability of multiple fires, that is, the expected output under stable conditions, which is used to quantify the deviation of the prediction; Indicates that 0 and the expression in brackets (i.e. ), if the expression in the brackets is a negative number, the function Replaced with 0 to prevent negative values ​​in the cumulative effect statistics; Preset advanced detection thresholds , if the cumulative effect statistic Exceeding the advanced detection threshold , then the current monitoring area is identified as having a fire risk; otherwise, the current monitoring area is identified as having a possible fire risk and further analysis by maintenance personnel is required; The fire identification situation of the multi-level verification mechanism is fed back to the automation control and emergency response module; the feedback information includes fire location information, fire occurrence time, fire confirmation status, fire source type speculation, etc.

[0012] Furthermore, the automation control and emergency response module classifies the fire danger level and builds an emergency response strategy library based on the fire identification situation, and then automatically selects the optimal emergency response strategy, including the following process: Classification of fire levels , the fire level is divided into four levels; among them, , and 1 represents the lowest level and 4 represents the highest level; the fire level can be calculated by weighting the fusion features of multi-modal fire data; Generate corresponding warning prompts according to the fire level ; Building an emergency response strategy library , where the emergency response strategy library contains emergency response strategies for different fire hazard levels, and each fire level corresponds to one or more optimal emergency response strategies; Based on the fire hazard classification and emergency response strategy library, the system selects the optimal emergency response strategy and adjusts the optimal emergency response strategy based on on-site conditions; among them, the adjustment function of the optimal emergency response strategy for: In the formula, It indicates the amount of adjustment to the emergency response strategy based on on-site conditions; It is understandable that on-site conditions are dynamically changing, including but not limited to the following aspects: personnel distribution, which refers to the location, number and status of personnel at the fire scene; equipment status, which refers to the availability and status of fire-fighting equipment (such as fire extinguishers, sprinkler systems, ventilation systems), and the operating conditions of non-fire-fighting equipment; building structure, which refers to the layout of the building, exit locations, fire partitions and other information. The building structure determines the path and speed of fire spread, affecting the choice of evacuation routes and the implementation of fire isolation measures.

[0013] Compared with the existing technology, the fire monitoring and identification system and method based on multi-sensor fusion provided by the present invention have the following advantages: The present invention integrates multiple sensors to collect multi-dimensional and multi-modal data, improves the comprehensiveness of the data, helps to understand the fire situation more comprehensively, uses data fusion processing algorithms to intelligently integrate information from different sensors, extracts multi-modal fire data fusion features, and reduces the false alarm rate of the overall system; by creating a three-dimensional fire monitoring model, and adjusting parameters and evaluating the performance of the created three-dimensional fire monitoring model, the accuracy of fire prediction is improved, so as to facilitate the subsequent adjustment of emergency response strategies; The present invention introduces real-time monitoring area parameters as input for system judgment, combined with the output of the fire probability assessment module, to predict the probability of fire in the current monitoring area in real time. The prediction results can provide important references for emergency response decisions, help decision makers quickly formulate response measures, and reduce fire risks. By setting up a multi-level verification mechanism for fire identification and providing feedback on fire identification, the false alarm rate can be reduced and the accuracy of fire identification can be improved. Each level of detection increases the confidence in fire confirmation and reduces the possibility of misjudgment, allowing the system to maintain a high degree of reliability when facing complex fire scenes. The present invention classifies fire hazard levels and constructs an emergency response strategy library according to fire identification conditions, and then automatically selects the optimal emergency response strategy. This rapid response capability helps to quickly control the spread of fire when a fire occurs. According to different fire levels and types, the system can select the most appropriate emergency response strategy to deal with it. At the same time, automated control reduces the interference of human factors on emergency response and improves the accuracy and reliability of the response.

[0014] To sum up, the present invention significantly improves the accuracy of fire monitoring and the efficiency of emergency response by integrating multiple sensors, realizing data fusion processing, three-dimensional model monitoring and evaluation, real-time prediction and multi-level detection, as well as automatic control and efficient emergency response. The various modules in the invention cooperate and work together to form an efficient and comprehensive fire monitoring and emergency response system, which realizes all-round monitoring of fires and timely and effective response. At the same time, it is of great significance to ensure personnel safety and reduce property losses, and ensure the efficient and stable operation of subsequent fire monitoring and identification systems based on multi-sensor fusion. BRIEF DESCRIPTION OF THE DRAWINGS

[0015] Figure 1 This is a module diagram of the fire monitoring and identification system based on multi-sensor fusion proposed in the present invention. DETAILED DESCRIPTION

[0016] The following will be combined with the drawings in the embodiments of the present invention to clearly and completely describe the technical solutions in the embodiments of the present invention. Obviously, the implementation regulations described are only part of the embodiments of the present invention, not all of the embodiments. Based on the embodiments of the present invention, all other embodiments obtained by ordinary technicians in this field without creative work are within the scope of protection of the present invention.

[0017] Reference Figure 1 , a fire monitoring and identification system based on multi-sensor fusion, which includes a multi-modal sensor data fusion processing module, a fire possibility intelligent assessment module, a fire probability real-time prediction module, a multi-level detection and feedback module, and an automatic control and emergency response module; The multimodal sensor data fusion processing module is used to collect raw data by integrating multiple sensors, and extract multimodal fire data fusion features in combination with data fusion processing algorithms; wherein the multiple sensors include but are not limited to temperature sensors, smoke sensors, flame detectors, infrared thermal imagers, etc.; The fire occurrence possibility intelligent assessment module is used to create a three-dimensional fire monitoring model based on the extracted multi-modal fire data fusion features, and to adjust parameters and evaluate the performance of the created three-dimensional fire monitoring model; The fire occurrence probability real-time prediction module is used to introduce real-time monitoring area parameters as input for system judgment, and predict the probability of fire occurrence in the current monitoring area state through the fire occurrence possibility assessment module; The multi-level detection and feedback module is used to set up a multi-level verification mechanism, identify fires through the multi-level verification mechanism, and provide feedback on fire identification. The automatic control and emergency response module is used to classify fire hazard levels and build an emergency response strategy library based on fire identification conditions, and then automatically select the optimal emergency response strategy.

[0018] It should be noted that the application object of the fire monitoring and identification system based on multi-sensor fusion in the embodiment of the present invention can be fire prevention and control in the field of public safety. Specifically, it can be to comprehensively analyze and process the collected multi-source sensor data through artificial intelligence and data analysis technology to achieve real-time monitoring and timely regulation of fire risks. Through data preprocessing and fusion analysis of the multimodal sensor data fusion processing module, model creation and evaluation of the fire possibility intelligent evaluation module, accurate prediction of the fire probability real-time prediction module, effective verification of the multi-level detection and feedback module, and optimal strategy selection of the automatic control and emergency response module, a highly integrated and intelligent fire monitoring and identification system is formed, so that the system can warn of fire risks in advance, respond to fire incidents in time, and ensure the safety of life and property of personnel; at the same time, the system also promotes the intelligent development of fire protection technology, meets the needs of efficient and accurate fire prevention and control, and realizes the transformation from passive response to active prevention. It is of great significance to improve the level of social public safety, optimize the allocation of emergency resources, and promote the construction of smart cities.

[0019] The multimodal sensor data fusion processing module collects raw data by integrating multiple sensors and extracts multimodal fire data fusion features by combining data fusion processing algorithms. The steps include: Step 101: Select corresponding sensor types and locations according to the characteristics of the monitoring area; wherein the characteristics of the monitoring area include the size of the space, the type of potential fire source, and the complexity of the environment; for example, increase the density of smoke sensors in areas with dense flammable materials, and deploy infrared thermal imagers in open areas to monitor distant fire sources; Step 102: Periodically obtain raw data from each sensor and mark it as ,in, Represents the sensor index variable, used to traverse the sensor; Step 103: Define the sensor set as ,in, Indicates the total number of sensors; Step 104: For each sensor , whose original data is represented as a time series ,in, Indicates time, Indicates sensor In time The total number of data points collected; Step 105: For each data point , set the data cleaning threshold To clean: In the formula, Represents the data point index variable, used to traverse the data points; Respectively represent sensors The mean and standard deviation of the data; Indicates invalid data; Step 106: Integrate the pre-processed multi-source sensor data, perform fusion processing on the acquired multi-source sensor data, and extract the multi-modal fire data fusion features; at the same time, dynamically allocate the weight of each sensor in the fusion process according to its accuracy and reliability; wherein the specific expression is: In the formula, Represents the multimodal fire data fusion feature, that is, a comprehensive feature vector containing information from multiple sensors, which is used to reflect the various characteristics of the fire, such as temperature, smoke concentration, the presence or absence of flames, etc. Represents the sensor index variable; Indicates the total number of sensors; It's time The weight of the dependency, indicating the sensor In time accuracy and reliability; Indicates that the slave sensor Functions that extract features from data; Respectively indicate adjustment of the sensor Coefficient of accuracy and reliability; Respectively represent sensors At current time The accuracy evaluation value and reliability evaluation value of

[0020] The fire occurrence possibility intelligent assessment module creates a three-dimensional fire monitoring model based on the extracted multi-modal fire data fusion features, and the steps of adjusting parameters and evaluating the performance of the created three-dimensional fire monitoring model include: Step 201: Obtain a multi-modal fire data fusion feature set: in, Represents the acquired multi-modal fire data fusion feature index; Indicates at time The multi-modal fire data fusion feature values ​​obtained above; Represents the number of multimodal fire data fusion features; Step 202: Define the set of environmental characteristics of the monitoring area: in, Indicates the number of environmental features in the monitored area; Step 203: Construct a three-dimensional fire monitoring model , where Represents the parameters of the three-dimensional fire monitoring model; Step 204: Through the loss function Measures the difference between the model prediction and the actual fire occurrence. Fire monitoring is a binary classification problem, that is, whether the fire occurs or not. The specific expression is; In the formula, Represents the total number of samples selected from the historical fire data; Indicates the fire status label, used to describe the time The fire status, where 0 means no fire and 1 means fire; represents the probability of fire occurrence predicted by the three-dimensional fire monitoring model, where ; Step 205: Update the three-dimensional fire monitoring model parameters using the gradient descent method , to minimize the loss function , where the specific expression is: In the formula, represents the updated three-dimensional fire monitoring model parameters; represents an iteration index, i.e., describes the number of iterations or stages in the training process of the three-dimensional fire monitoring model, and is used to iteratively update the parameters of the three-dimensional fire monitoring model during the training process; Indicates Parameters of the three-dimensional fire monitoring model for the first iteration; Represents the learning rate, which is used to control the step size of parameter adjustment; It means that the loss function is The gradient at Step 206: Use a cross-validation method to evaluate the performance of the three-dimensional fire monitoring model: A1. Extract a complete dataset containing multimodal fire data fusion features and fire status labels from historical data , where the dataset Each record in Including multi-modal fire data fusion feature set and the corresponding fire status label ; In step A1, the complete dataset It collects the multimodal fire data fusion feature set and the corresponding fire status label, that is, each piece of data not only contains the multimodal fire data fusion feature of the monitoring area, but also contains the label information of whether a fire has occurred, and 0 means no fire and 1 means fire; A2. Dataset Randomly divide into Mutually exclusive subsets ; In step A2, The value is 5 or 10 and is called Fold cross validation; each subset contains the same amount of data, and each subset should maintain the representativeness of the data, that is, contain samples with and without fire conditions; A3. For each subset , which is used as the validation set, and the rest The subsets are used as training sets to train the three-dimensional fire monitoring model; among them, represents the index of the subset, and ; A4. During the training of the 3D fire monitoring model, the training set data is used to learn the model parameters, and the validation set data is used to learn the model parameters. Test the performance of the model on A5. Repeat steps A3-A4 until each subset is used as a validation set. A6. After each verification, a set of performance indicators is obtained. The performance indicators obtained from the verification are averaged to obtain the performance of the three-dimensional fire monitoring model in the data set. Comprehensive performance evaluation on For example, taking the calculation of the accuracy of the three-dimensional fire monitoring model as an example, assuming Respectively expressed in The performance index obtained in the first verification is the comprehensive performance index of the three-dimensional fire monitoring model. It is expressed as: , where No. The performance indicators obtained by verification; In step 206, based on the performance indicators obtained by cross-validation, the model parameters can be further optimized or the model structure can be adjusted. At the same time, the performance of the three-dimensional fire monitoring model can be evaluated more objectively and comprehensively to improve the generalization ability and prediction accuracy of the model.

[0021] The fire occurrence probability real-time prediction module introduces real-time monitoring area parameters as input for system determination. The steps of predicting the probability of fire occurrence in the current monitoring area state through the fire occurrence possibility assessment module include: Step 301: For at the moment The real-time monitoring area index vector includes the current multi-modal fire data fusion characteristics and the environmental characteristics of the current monitoring area; In step 301, the real-time monitoring area index vector includes the current multi-modal fire data fusion features and the environmental features of the current monitoring area; Step 302: Real-time monitoring area index vector Input the fire probability assessment module to predict the probability of fire in the current monitoring area, and then obtain the real-time prediction result of fire probability. .

[0022] The multi-level detection and feedback module sets up a multi-level verification mechanism, and the steps of performing fire identification through the multi-level verification mechanism and feeding back the fire identification situation include: Step 401: Setting up a multi-level verification mechanism, including primary detection, intermediate analysis, and advanced confirmation; Step 402, the steps of performing fire identification through a multi-level verification mechanism are as follows: Preset primary detection threshold , if the fire probability is predicted in real time The value exceeds the primary detection threshold , then enter the intermediate analysis; otherwise, identify that there is no fire risk in the current monitoring area; Real-time prediction of results by calculating the probability of fire Rate of change : In the formula, Indicates at time The real-time prediction result value of the probability of fire occurrence; Preset intermediate detection threshold ,like Continues to increase and exceeds the intermediate detection threshold , then enter advanced confirmation; otherwise, identify the current monitoring area as a possible fire risk, requiring further analysis by maintenance personnel; Assume that the Moment Real-time prediction results of multiple fire occurrence probabilities and calculation of their cumulative effect statistics : In the formula, Indicates Moment The cumulative effect statistics of the real-time prediction results of multiple fire occurrence probabilities are used to monitor the cumulative changes in the process; Indicates Moment The cumulative effect statistics of the real-time prediction results of multiple fire occurrence probabilities provide the historical background of the cumulative effect, making the current moment The calculation of the cumulative effect takes into account all previous changes; Indicates at time Real-time prediction result value of multiple fire occurrence probability; It represents the benchmark value of the real-time prediction of the probability of multiple fires, that is, the expected output under stable conditions, which is used to quantify the deviation of the prediction; Indicates that 0 and the expression in brackets (i.e. ), if the expression in the brackets is a negative number, the function Replaced with 0 to prevent negative values ​​in the cumulative effect statistics; Preset advanced detection thresholds , if the cumulative effect statistic Exceeding the advanced detection threshold , then the current monitoring area is identified as having a fire risk; otherwise, the current monitoring area is identified as having a possible fire risk and further analysis by maintenance personnel is required; Step 403: Feedback the fire identification situation of the multi-level verification mechanism to the automation control and emergency response module; In step 403, the feedback information includes fire location information, fire occurrence time, fire confirmation status, fire source type estimation, etc.

[0023] The steps of the automatic control and emergency response module to classify fire danger levels and build an emergency response strategy library based on fire identification, and then automatically select the optimal emergency response strategy include: Step 501: Classify fire levels , the fire level is divided into four levels; among them, , and 1 represents the lowest level and 4 represents the highest level; the fire level can be calculated by weighting the fusion features of multi-modal fire data; Step 502: Generate corresponding warning prompts according to the classified fire levels ; Step 503: Build an emergency response strategy library , where the emergency response strategy library contains emergency response strategies for different fire hazard levels, and each fire level corresponds to one or more optimal emergency response strategies; Step 504: Based on the fire risk classification and the emergency response strategy library, the system selects the optimal emergency response strategy and adjusts the optimal emergency response strategy in combination with the on-site conditions; wherein the adjustment function of the optimal emergency response strategy for: In the formula, It indicates the amount of adjustment to the emergency response strategy based on on-site conditions; In step 504, the on-site conditions are dynamically changing, including but not limited to the following aspects: personnel distribution, i.e., the location, number and status of personnel in the fire scene; equipment status, i.e., the availability and status of fire-fighting equipment (such as fire extinguishers, sprinkler systems, ventilation systems), and the operating conditions of non-fire-fighting equipment; building structure, i.e., information such as the layout of the building, exit locations, and fire partitions. The building structure determines the path and speed of fire spread, affecting the selection of evacuation routes and the implementation of fire isolation measures.

[0024] In the embodiment of the present invention, by collecting multi-dimensional data from different sensors (such as temperature sensors, smoke sensors, flame detectors, infrared thermal imagers, etc.), the data of multiple sensors are complementary, and different stages and characteristics of fire can be more accurately identified. At the same time, a single sensor may affect the result due to failure or false alarm, and multimodal fusion can reduce this risk; by using algorithms to process raw data, multimodal fire data fusion features are extracted, and these features can more comprehensively reflect the fire situation; by constructing a three-dimensional model based on the fused multimodal data to intuitively display the fire situation, and optimizing the parameters of the model to ensure its accuracy and response speed, the model parameters are adjustable and can adapt to the monitoring needs of different environments and fire types; by introducing real-time monitoring parameters, combined with the output of the fire occurrence possibility assessment module, the probability of fire in the current monitoring area is calculated in real time to improve the accuracy of the prediction; by different levels of detection algorithms and rules, the fire is verified multiple times, and the multi-layer verification mechanism greatly reduces the false alarm rate and improves the reliability of the system; by dividing the fire hazard level, building an emergency response strategy library, according to the fire level and actual situation, automatically select and execute the most appropriate emergency response strategy, reasonably allocate emergency resources, reduce waste, and improve rescue efficiency. In summary, the examples of the present invention solve the problems of incomplete monitoring and insufficient control in current fire prevention and management. In actual situations, more data and contextual information may be needed to make specific decisions and optimize solutions.

[0025] In addition, the formulas involved in the above are all calculated by removing dimensions and taking their numerical values. They are a formula that is closest to the actual situation obtained by collecting a large amount of data and performing software simulation. The proportional coefficient in the formula and the various preset thresholds in the analysis process are set by technical personnel in this field according to actual conditions or obtained by simulating a large amount of data; the size of the proportional coefficient is to quantify each parameter to obtain a specific value for subsequent comparison. The size of the proportional coefficient depends on the amount of sample data and the preliminary setting of the corresponding processing coefficient for each group of sample data by technical personnel in this field; as long as it does not affect the proportional relationship between the parameter and the quantized value.

[0026] Each embodiment in this specification is described in a progressive manner, and the same or similar parts between the embodiments can be referred to each other, and each embodiment focuses on the differences from other embodiments. In particular, for the device embodiment, since it is basically based on the method embodiment, the description is relatively simple, and the relevant parts can be referred to the partial description of the method embodiment.

[0027] For the convenience of description, the above device is described in various units according to their functions. Of course, when implementing the present application, the functions of each unit can be implemented in the same or multiple software and / or hardware.

[0028] It will be appreciated by those skilled in the art that embodiments of the present invention may be provided as methods, systems, or computer program products. Therefore, the present invention may take the form of a complete hardware embodiment, a complete software embodiment, or an embodiment combining software and hardware. Furthermore, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media (including but not limited to disk storage, CD-ROM, optical storage, etc.) containing computer-usable program code.

[0029] The present invention is described with reference to flowcharts and / or block diagrams of methods, devices (systems), and computer program products according to embodiments of the present invention. It should be understood that each process and / or block in the flowchart and / or block diagram, as well as the combination of processes and / or blocks in the flowchart and / or block diagram, can be implemented by computer program instructions. These computer program instructions can be provided to a processor of a general-purpose computer, a special-purpose computer, an embedded processor, or other programmable data processing device to generate a machine, so that the instructions executed by the processor of the computer or other programmable data processing device generate instructions for implementing the processes in the flowchart and / or block diagram. Figure 1 A process or multiple processes and / or boxes Figure 1 A device that provides the functions specified in a block or multiple blocks.

[0030] These computer program instructions may also be stored in a computer-readable memory capable of directing a computer or other programmable data processing device to operate in a specific manner, so that the instructions stored in the computer-readable memory produce an article of manufacture comprising an instruction device, which implements the process Figure 1 A process or multiple processes and / or boxes Figure 1 A function specified in one or more boxes.

[0031] These computer program instructions can also be loaded onto a computer or other programmable data processing device so that a series of operating steps are executed on the computer or other programmable device to produce a computer-implemented process, thereby providing instructions for implementing the process. Figure 1A process or multiple processes and / or boxes Figure 1 The steps for the functions specified in one or more boxes.

[0032] Secondly: In the drawings of the embodiments disclosed in the present invention, only the structures related to the embodiments disclosed in the present invention are involved, and other structures can refer to the general design. In the absence of conflict, the same embodiment and different embodiments of the present invention can be combined with each other; Finally: The above description is only a preferred specific implementation manner of the present invention, but the protection scope of the present invention is not limited thereto. Any technician familiar with the technical field can make equivalent replacements or changes according to the technical scheme and inventive concept of the present invention within the technical scope disclosed by the present invention, which should be covered by the protection scope of the present invention.

Claims

1. Fire monitoring and identification system based on multi-sensor fusion, characterized by: It includes multi-modal sensor data fusion processing module, fire probability intelligent assessment module, fire probability real-time prediction module, multi-level detection and feedback module, and automatic control and emergency response module; The multimodal sensor data fusion processing module is used to collect raw data by integrating multiple sensors and extract multimodal fire data fusion features in combination with data fusion processing algorithms; The fire occurrence possibility intelligent assessment module is used to create a three-dimensional fire monitoring model based on the extracted multi-modal fire data fusion features, and to adjust parameters and evaluate the performance of the created three-dimensional fire monitoring model; The fire occurrence probability real-time prediction module is used to introduce real-time monitoring area parameters as input for system judgment, and predict the probability of fire occurrence in the current monitoring area state through the fire occurrence possibility assessment module; The multi-level detection and feedback module is used to set up a multi-level verification mechanism, identify fires through the multi-level verification mechanism, and provide feedback on fire identification. The automatic control and emergency response module is used to classify fire hazard levels and build an emergency response strategy library based on fire identification conditions, and then automatically select the optimal emergency response strategy.

2. The fire monitoring and identification system based on multi-sensor fusion according to claim 1 is characterized in that: The multimodal sensor data fusion processing module collects raw data by integrating multiple sensors and extracts multimodal fire data fusion features by combining data fusion processing algorithms. The process includes: Select the corresponding sensor type and location according to the characteristics of the monitoring area; the characteristics of the monitoring area include the size of the space, the type of potential fire source, and the complexity of the environment; Raw data is periodically obtained from each sensor and labeled ,in, Represents the sensor index variable, used to traverse the sensor; Define the sensor set as ,in, Indicates the total number of sensors; For each sensor , whose original data is represented as a time series ,in, Indicates time, Indicates sensor In time The total number of data points collected; For each data point , set the data cleaning threshold To clean: In the formula, Represents the data point index variable, used to traverse the data points; Respectively represent sensors The mean and standard deviation of the data; Indicates invalid data; Integrate the preprocessed multi-source sensor data, perform fusion processing on the acquired multi-source sensor data, and extract the multi-modal fire data fusion features; at the same time, according to the accuracy and reliability of each sensor, dynamically allocate its weight in the fusion process; the specific expression is: In the formula, Represents the fusion characteristics of multi-modal fire data; Represents the sensor index variable; Indicates the total number of sensors; It's time The weight of the dependency, indicating the sensor In time accuracy and reliability; Indicates that the slave sensor Functions that extract features from data; Respectively indicate adjustment of the sensor Coefficient of accuracy and reliability; Respectively represent sensors At current time The accuracy evaluation value and reliability evaluation value of 3. The fire monitoring and identification system based on multi-sensor fusion according to claim 1 is characterized in that: The fire occurrence possibility intelligent assessment module creates a three-dimensional fire monitoring model based on the extracted multi-modal fire data fusion features, and the process of parameter adjustment and performance evaluation of the created three-dimensional fire monitoring model includes: Get the multimodal fire data fusion feature set: in, Represents the acquired multi-modal fire data fusion feature index; Indicates at time The multi-modal fire data fusion feature values ​​obtained above; Represents the number of multimodal fire data fusion features; Define the set of environmental characteristics of the monitoring area: in, Indicates the number of environmental features in the monitored area; Building a 3D fire monitoring model , where Represents the parameters of the three-dimensional fire monitoring model; Through the loss function Measures the difference between the model prediction and the actual fire occurrence, where the specific expression is; In the formula, Represents the total number of samples selected from the historical fire data; Indicates the fire status label, used to describe the time The fire status, where 0 means no fire and 1 means fire; represents the probability of fire occurrence predicted by the three-dimensional fire monitoring model, where ; Updating parameters of 3D fire monitoring model using gradient descent method , to minimize the loss function , where the specific expression is: In the formula, represents the updated three-dimensional fire monitoring model parameters; represents an iteration index, i.e., describes the number of iterations or stages in the training process of the three-dimensional fire monitoring model, and is used to iteratively update the parameters of the three-dimensional fire monitoring model during the training process; Indicates Parameters of the three-dimensional fire monitoring model for the first iteration; Represents the learning rate, which is used to control the step size of parameter adjustment; It means that the loss function is The gradient at The cross-validation method was used to evaluate the performance of the three-dimensional fire monitoring model.

4. The fire monitoring and identification system based on multi-sensor fusion according to claim 3 is characterized in that: The fire probability intelligent assessment module uses a cross-validation method to evaluate the performance of the three-dimensional fire monitoring model. The process includes: A1. Extract a complete dataset containing multimodal fire data fusion features and fire status labels from historical data , where the dataset Each record in Including multi-modal fire data fusion feature set and the corresponding fire status label ; A2. Dataset Randomly divide into Mutually exclusive subsets ; A3. For each subset , and use it as a validation set. The subsets are used as training sets to train the three-dimensional fire monitoring model; among them, represents the index of the subset, and ; A4. During the training of the 3D fire monitoring model, the training set data is used to learn the model parameters, and the validation set data is used to learn the model parameters. Test the performance of the model on A5. Repeat steps A3-A4 until each subset is used as a validation set. A6. After each verification, a set of performance indicators is obtained. The performance indicators obtained from the verification are averaged to obtain the performance of the three-dimensional fire monitoring model in the data set. Comprehensive performance evaluation on .

5. The fire monitoring and identification system based on multi-sensor fusion according to claim 1 is characterized in that: The fire probability real-time prediction module introduces real-time monitoring area parameters as input for system judgment. The process of predicting the probability of fire in the current monitoring area state through the fire probability assessment module includes: make For at the moment The real-time monitoring area index vector includes the current multi-modal fire data fusion characteristics and the environmental characteristics of the current monitoring area; Real-time monitoring of regional indicator vectors Input the fire probability assessment module to predict the probability of fire in the current monitoring area, and then obtain the real-time prediction result of fire probability. .

6. The fire monitoring and identification system based on multi-sensor fusion according to claim 1 is characterized in that: The multi-level detection and feedback module sets up a multi-level verification mechanism. The process of fire identification and feedback of fire identification status through the multi-level verification mechanism includes: Set up a multi-level verification mechanism, including primary testing, intermediate analysis and advanced confirmation; The process of fire identification through a multi-level verification mechanism is as follows: Preset primary detection threshold , if the fire probability is predicted in real time The value exceeds the primary detection threshold , then enter the intermediate analysis; otherwise, identify that there is no fire risk in the current monitoring area; Real-time prediction of results by calculating the probability of fire Rate of change : In the formula, Indicates at time The real-time prediction result value of the probability of fire occurrence; Preset intermediate detection threshold ,like Continues to increase and exceeds the intermediate detection threshold , then enter advanced confirmation; otherwise, identify the current monitoring area as a possible fire risk, requiring further analysis by maintenance personnel; Assume that you select Moment Real-time prediction results of multiple fire occurrence probabilities and calculation of their cumulative effect statistics : In the formula, Indicates Moment The cumulative effect statistics of the real-time prediction results of multiple fire occurrence probabilities are used to monitor the cumulative changes in the process; Indicates Moment The cumulative effect statistics of the real-time prediction results of multiple fire occurrence probabilities; Indicates at time Real-time prediction result value of multiple fire occurrence probability; It represents the benchmark value for real-time prediction of the probability of multiple fires; Indicates that 0 and the expression in brackets (i.e. ) is the maximum value between Preset advanced detection thresholds , if the cumulative effect statistic Exceeding the advanced detection threshold , then the current monitoring area is identified as having a fire risk; otherwise, the current monitoring area is identified as having a possible fire risk and further analysis by maintenance personnel is required; The fire identification situation of the multi-level verification mechanism is fed back to the automation control and emergency response module.

7. The fire monitoring and identification system based on multi-sensor fusion according to claim 1 is characterized in that: The automation control and emergency response module classifies fire danger levels and builds an emergency response strategy library based on fire identification, and then automatically selects the optimal emergency response strategy. The process includes: Classification of fire levels , the fire level is divided into four levels; among them, , with 1 representing the lowest level and 4 representing the highest level; Generate corresponding warning prompts according to the fire level ; Building an emergency response strategy library , where the emergency response strategy library contains emergency response strategies for different fire hazard levels, and each fire level corresponds to one or more optimal emergency response strategies; Based on the fire hazard classification and emergency response strategy library, the system selects the optimal emergency response strategy and adjusts the optimal emergency response strategy based on on-site conditions; among them, the adjustment function of the optimal emergency response strategy for: In the formula, Represents the amount of adjustment to the emergency response strategy based on on-site conditions.

Images