Construction fire detection method and system based on Internet of Things

Abstract

The invention discloses a construction fire detection method and system based on the Internet of Things, which relates to the technical field of electrical digital data processing, and specifically comprises: dividing a construction site into regions and grading regional sensitivity according to manual labeling results; grouping node clusters and calculating node energy coefficients according to environmental data collected by a wireless sensor network, and completing the election of leading nodes by sorting the energy coefficients; recording a fixed monitoring area to obtain video data and sending the video data in real time; establishing an environmental data set, calculating a confidence level probability model according to a D-S model, and performing probability screening through a confidence interval to obtain a fire judgment result; processing video data and extracting image features through a frame difference method, and obtaining a fire judgment result according to a back propagation neural network and a fixed threshold detection method; and performing on-site sound and light alarms, cloud platform alarm reminders, and sending notifications to fire departments based on the fire judgment results, so as to complete fire detection and early warning.

Description

Technical Field

[0001] The present invention relates to the technical field of electronic digital data processing, and in particular to a construction fire detection method and system based on the Internet of Things. Background Art

[0002] The construction industry is characterized by long engineering cycles, complex construction links, and high environmental risks, which can easily breed various fire hazards during the construction and production process. Due to the large number of construction workers and their wide range of sources, a considerable number of them have not received systematic fire protection knowledge training, lack safety awareness, and have incomplete knowledge of fire protection and self-rescue. In addition, the construction process requires a large number of hot work operations such as welding and cutting, which can easily cause fires due to inadequate protective measures. In addition, some of the power grid lines used in the construction are temporarily laid, which can easily cause circuit overload and short circuit problems due to improper use of electricity by construction workers, causing the lines themselves and surrounding equipment to catch fire. In addition, a large number of combustible building materials are required during the construction process, and improper management problems such as the failure to set up special warehouses for flammable and explosive items and the failure to store them separately also increase the risk of fire.

[0003] At present, the Chinese invention patent with application number CN202110206018.7 discloses a fire protection facility detection method, device and medium based on the Internet of Things. The application scheme includes a fire protection facility detection terminal, an Internet of Things mobile terminal and a background monitoring terminal, and the Internet of Things mobile terminal communicates with the fire protection facility detection terminal and the background monitoring terminal respectively; the fire protection facility detection terminal includes multiple detection sensor modules for obtaining detection data of fire protection facilities; the Internet of Things mobile terminal includes a receiving module for receiving detection data and a sending module for sending detection data; the background monitoring terminal includes: a data processing module for processing detection data; a data storage module for storing detection data after data processing; a data analysis module for analyzing the detection data after data processing and obtaining analysis results; a result display module for displaying analysis results. Although the application has the effect of improving the efficiency of fire protection facility overall detection, the environment of the construction site is complex, and there are a large number of open fire operations, which can easily cause misidentification and false alarms of fire conditions. At the same time, as the construction progress increases, the application attributes and fire risks of various areas of the construction site are constantly changing. For example, the welding operation area becomes a flammable material storage area after completion. Therefore, the fire detection system with fixed sensitivity is prone to missed detection and false alarm problems. In addition, when the traditional IoT-based fire detection system is applied to large-scale outdoor construction sites, there are problems of high energy consumption and limited working time. Summary of the invention

[0004] The technical problem solved by the present invention is that traditional fire detection methods and systems are not suitable for the complex environment of construction sites and the rapid changes in fire risks, are prone to false fire alarms and missed detections, and have the problem of high energy consumption of the detection system, and have a limited scope of use.

[0005] In order to solve the above technical problems, the present invention provides the following technical solutions:

[0006] The construction fire detection method based on the Internet of Things specifically includes the following steps:

[0007] Step S1, dividing the construction site into regions and grading the regional sensitivity through manual labeling, and sending the construction area classification results and sensitivity grading results to the data processing center subsystem;

[0008] Step S2, collecting environmental data through a wireless sensor network, clustering sensor nodes according to the construction area classification result, completing the leader node election according to the calculated energy coefficient of the nodes in the cluster, and sending the collected environmental data to the data processing center subsystem through the leader node;

[0009] Step S3, recording the fixed monitoring area to obtain video data, and sending the video data to the data processing center subsystem in real time;

[0010] Step S4, establishing an environmental data set based on various sensor data collected by the wireless sensor network, calculating a confidence level probability model based on DS theory and the environmental data set, and performing probability screening according to the confidence interval combined with the sensitivity level to obtain a fire judgment result;

[0011] Step S5, processing the video data according to the frame difference method to obtain a suspicious area image, extracting the flame features and smoke features of the suspicious area image, and obtaining a fire judgment result through a back propagation neural network and a fixed threshold detection method;

[0012] Step S6, sending a fire warning based on the fire judgment result obtained, including sending out sound and light alarms at the construction site and broadcasting fire notifications on the cloud platform, completing fire detection and early warning.

[0013] As a preferred solution of the construction fire detection method based on the Internet of Things described in the present invention, the construction site is spatially classified on the cloud platform, and the construction site is divided into a flammable material area, an open fire operation area and a general construction area by manual marking;

[0014] Time classification is carried out through clock-in operation based on the cloud platform. Construction personnel enter the work mode by clocking in and out. Open flame operation areas are classified as low-sensitivity areas, ordinary construction areas are classified as medium-sensitivity areas, and flammable material areas are classified as high-sensitivity areas.

[0015] The construction workers enter the off-duty mode by clocking in and out, and the entire construction site area is classified as a high-sensitivity area. By dividing the construction area into low-sensitivity area, medium-sensitivity area and high-sensitivity area, when analyzing and processing the obtained sensor data and video data in the data processing center subsystem, the weight values ​​corresponding to the low-sensitivity area, medium-sensitivity area and high-sensitivity area are flexibly introduced in the corresponding calculation process to improve the accuracy of the fire judgment results.

[0016] As a preferred solution of the Internet of Things-based construction fire detection method described in the present invention, the temperature data, carbon monoxide concentration data and ultraviolet light intensity data of the construction area are collected through the wireless sensor network, the sensor nodes are clustered according to the construction area classification results, and the environmental data are sent to the data processing center subsystem through the leader node in the cluster.

[0017] As a preferred solution of the construction fire detection method based on the Internet of Things described in the present invention, the wireless sensor network realizes leap-layer data transmission by adding an inter-layer interface between the network layer, MAC layer and physical layer in the routing protocol, and regularly conducts leadership node election by calculating the energy coefficient, and the calculation expression is:

[0018] ;

[0019] Where U represents the energy coefficient, Indicates the maximum power of the node. Indicates the remaining power of the candidate node, represents the paradigm distance between the candidate node and other nodes in the cluster, It represents the cumulative sum of the normalized distances from all nodes in the cluster to the communication base station.

[0020] As a preferred solution of the construction fire detection method based on the Internet of Things described in the present invention, the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network are used to establish an environmental data set, and the confidence level probability model is calculated for the environmental data set by the DS theory, and the confidence level is initialized and assigned to the set corresponding to the temperature data, carbon monoxide concentration data and ultraviolet light intensity data, and then the confidence level is redistributed according to the environmental data newly collected by the wireless sensor network, and the confidence level of each data set is continuously corrected as the collected data accumulates, and all data sets are traversed, and probability screening is performed by combining the confidence interval of the sensitivity level and the fire judgment result is output. The calculation expression is:

[0021] ;

[0022] in, represents the trust function that the data set should respond to, represents the likelihood function corresponding to the data set, represents the exact probability, Indicates the sensitivity weight corresponding to the sensor position. In the specific calculation, when the data source area is a low-sensitivity area, the sensitivity weight is used When the data source area is the medium sensitive area, the sensitivity weight is When the data source area is a high-sensitivity area, the sensitivity weight is .

[0023] As a preferred solution of the construction fire detection method based on the Internet of Things described in the present invention, the video is sampled and segmented by a frame difference method, the image of the suspicious area is obtained by subtracting adjacent frame images, the flame feature and smoke feature of the suspicious area are extracted, the flame feature vector is obtained by calculating the image information entropy and the brightness change rate of the suspicious area, the smoke feature vector is obtained by calculating the area growth rate and the centroid deviation of the suspicious area, the extracted flame feature vector and smoke feature vector are input into a trained back propagation neural network to obtain an output prediction value, and the fixed threshold detection method is used to obtain the fire judgment result, and its calculation expression is:

[0024] ;

[0025] in, represents image information entropy, k represents grayscale value, G represents the maximum grayscale value, represents the probability that the gray value of a pixel is k, ln represents the natural logarithm, represents the predicted value, Indicates the fire existence judgment, Indicates the fixed threshold value of the image. In the specific calculation, when the data source area is a low-sensitivity area, the fixed threshold value is used When the data source area is the medium sensitive area, the fixed threshold value is used When the data source area is a high-sensitivity area, the fixed threshold value is used , obtained When the output result is 1, it is judged that there is a fire in the data source area, and the obtained When the output result is 0, it is judged that there is no fire in the data source area.

[0026] As a preferred solution of the Internet of Things-based construction fire detection method described in the present invention, the fire judgment result is received through the terminal subsystem, a fire alarm is issued at the construction site through the sound and light alarm module, and a notification is sent to the fire department through the cloud platform.

[0027] The construction fire detection system based on the Internet of Things includes: a grading subsystem, a wireless sensor network monitoring subsystem, an optical monitoring subsystem, a data processing center subsystem and a terminal subsystem;

[0028] The grading subsystem includes a punch-in module, an area division module and a sensitivity grading module, which are used to divide the construction site into areas and grade the regional sensitivity through manual labeling, and send the construction area classification results and sensitivity grading results to the data processing center subsystem. The area division module and the sensitivity grading module are connected to the data processing center subsystem, and the grading subsystem is connected to the data processing center subsystem;

[0029] The wireless sensor network detection subsystem includes a hybrid sensor network module, a leader node self-checking module and a wireless communication module, which are used to collect environmental data through the wireless sensor network, cluster the sensor nodes according to the construction area classification result, complete the leader node election according to the calculated energy coefficient of the nodes in the cluster, and send the collected environmental data to the data processing center subsystem through the leader node. The wireless sensor network detection subsystem is connected to the data processing center subsystem;

[0030] The optical detection subsystem includes a video acquisition module, a video storage module and a data transmission module, which are used to record a fixed monitoring area to obtain video data and send the video data to the data processing center subsystem in real time; the optical detection subsystem is connected to the data processing center subsystem;

[0031] The data processing center subsystem includes a data aggregation module, a wireless sensor detection data processing module, an optical detection data processing module and a fire judgment module, which are used to establish an environmental data set based on various sensor data collected by the wireless sensor network, calculate a confidence level probability model based on the DS theory and the environmental data set, perform probability screening based on the confidence interval combined with the sensitivity level to obtain a fire judgment result, process the video data based on the frame difference method to obtain a suspicious area image, extract the flame features and smoke features of the suspicious area image, and obtain the fire judgment result through the back propagation neural network and fixed threshold detection method;

[0032] The terminal subsystem includes a cloud platform module, an audible and visual alarm module, and a fire alarm module, which are used to send fire warnings based on the fire judgment results, including sending audible and visual alarms at the construction site and broadcasting fire notifications on the cloud platform, completing fire detection and early warning;

[0033] Wherein, a data set is established according to the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network, the sensor nodes are clustered according to the construction area classification result, and the temperature data, carbon monoxide concentration data and ultraviolet light intensity data are sent to the data processing center subsystem through the leader node in the cluster;

[0034] An inter-layer interface is added between the network layer, MAC layer and physical layer in the routing protocol of the wireless sensor network to realize leap-layer data transmission. The energy coefficients of the nodes in the cluster are calculated regularly, the energy coefficients are sorted, and the node with the largest energy coefficient is selected as the leader node. The calculation expression is:

[0035] ;

[0036] Where U represents the energy coefficient, Indicates the maximum power of the node. Indicates the remaining power of the candidate node, represents the paradigm distance between the candidate node and other nodes in the cluster, Represents the cumulative sum of the paradigm distances from all nodes in the cluster to the communication base station;

[0037] An environmental data set is established based on the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network. The confidence level probability model is calculated for the environmental data set using the DS model. Probability screening is performed based on the confidence interval combined with the sensitivity level and the fire judgment result is output. The calculation expression is:

[0038] ;

[0039] in, represents the trust function corresponding to the environmental dataset, represents the likelihood function corresponding to the data set, represents the exact probability, Indicates the sensitivity weight corresponding to the sensor position;

[0040] The video is sampled and segmented according to the frame difference method. The image of the suspicious area is obtained by subtracting adjacent frame images. The flame feature and smoke feature of the suspicious area are extracted. The image information entropy and brightness change rate of the suspicious area are calculated to obtain the flame feature vector. The area growth rate and centroid offset of the suspicious area are calculated to obtain the smoke feature vector. The extracted flame feature vector and smoke feature vector are input into the trained back propagation neural network. The fire judgment result is obtained by the fixed threshold detection method. The calculation expression is:

[0041] ;

[0042] in, represents image information entropy, k represents grayscale value, G represents the maximum grayscale value, represents the probability that the gray value of a pixel is k, ln represents the natural logarithm, represents the predicted value, Indicates the fire existence judgment, Indicates the fixed threshold value of the image.

[0043] As a preferred solution of the construction fire detection system based on the Internet of Things described in the present invention, wherein: the classification subsystem classifies the construction areas and grades the sensitivity of the construction site according to time and space, and sends the construction area classification results and sensitivity grading results to the data processing center subsystem;

[0044] The wireless sensor network detection subsystem collects detection environment data, clusters the sensor nodes according to the construction area classification results, and the leader node self-inspection module periodically conducts leader node elections in each cluster, and the leader node sends the collected data to the data processing center subsystem;

[0045] The video acquisition module in the optical monitoring subsystem records a fixed area and sends the video data to the data processing center subsystem in real time;

[0046] The data processing center subsystem aggregates, analyzes and processes data. The wireless sensor data processing module establishes a confidence probability model, performs fusion calculations on sensor data, calculates confidence intervals based on sensitivity grading results, outputs fire judgment results based on sensor network data, extracts features from images through the optical monitoring data processing module, obtains fire judgment results through the trained back-propagation neural network, and sends the results to the terminal subsystem.

[0047] The cloud platform operation module of the terminal subsystem is used to clock in construction personnel and mark construction areas, and the early warning module is used to issue on-site sound and light alarms and remote message notifications from the cloud platform.

[0048] The beneficial effects of the present invention are as follows: by having construction personnel clock in and out and manually mark areas every day, the construction areas are classified and the regional sensitivity is graded. Different weights are introduced in the subsequent data processing process according to the sensitivity grade, which is conducive to adapting to the complex environment and fast-changing characteristics of the construction site, improving the detection adaptability while reducing the false alarm rate of fire alarms. By adding a leap-layer communication function and a self-inspection mechanism for high-energy-consuming leader nodes to the wireless sensor network, it is conducive to ensuring that the sensor reduces overall power consumption on the basis of effective operation and prolongs the operating time of the sensor network. The use of carbon monoxide and ultraviolet sensors as a supplement to temperature sensors is conducive to reducing environmental interference in construction operations and improving the accuracy of collected environmental data. The recognition of flame and smoke characteristics of images by neural networks is conducive to real-time and efficient detection of small fire sources and elimination of fire hazards in the early stages of development. BRIEF DESCRIPTION OF THE DRAWINGS

[0049] Figure 1 A schematic diagram of the basic process of a construction fire detection method based on the Internet of Things provided by an embodiment of the present invention;

[0050] Figure 2 A schematic diagram of a building construction fire detection system based on the Internet of Things provided by one embodiment of the present invention. DETAILED DESCRIPTION

[0051] In order to make the above-mentioned objects, features and advantages of the present invention more obvious and easy to understand, the specific implementation methods of the present invention are described in detail below in conjunction with the drawings. It is obvious that the described embodiments are only part of the embodiments of the present invention, but not all of the embodiments.

[0052] Example 1, reference Figure 1 , is an embodiment of the present invention, and provides a construction fire detection method based on the Internet of Things, comprising:

[0053] Step S1, dividing the construction site into regions and grading the regional sensitivity through manual labeling, and sending the construction area classification results and sensitivity grading results to the data processing center subsystem;

[0054] Step S2, collecting environmental data through a wireless sensor network, clustering sensor nodes according to the construction area classification result, completing the leader node election according to the calculated energy coefficient of the nodes in the cluster, and sending the collected environmental data to the data processing center subsystem through the leader node;

[0055] Step S3, recording the fixed monitoring area to obtain video data, and sending the video data to the data processing center subsystem in real time;

[0056] Step S4, establishing an environmental data set based on various sensor data collected by the wireless sensor network, calculating a confidence level probability model based on DS theory and the environmental data set, and performing probability screening according to the confidence interval combined with the sensitivity level to obtain a fire judgment result;

[0057] Step S5, processing the video data according to the frame difference method to obtain a suspicious area image, extracting the flame features and smoke features of the suspicious area image, and obtaining a fire judgment result through a back propagation neural network and a fixed threshold detection method;

[0058] Step S6, sending a fire warning based on the fire judgment result obtained, including sending out sound and light alarms at the construction site and broadcasting fire notifications on the cloud platform, completing fire detection and early warning.

[0059] In this embodiment, the construction site is spatially classified on the cloud platform, and the construction site is divided into a flammable material area, an open fire operation area, and a general construction area through manual marking;

[0060] Time classification is carried out through clock-in operation based on the cloud platform. Construction personnel enter the work mode by clocking in and out. Open flame operation areas are classified as low-sensitivity areas, ordinary construction areas are classified as medium-sensitivity areas, and flammable material areas are classified as high-sensitivity areas.

[0061] The construction workers enter the off-duty mode by clocking in and out, and the entire construction site area is classified as a high-sensitivity area. By dividing the construction area into low-sensitivity area, medium-sensitivity area and high-sensitivity area, when analyzing and processing the obtained sensor data and video data in the data processing center subsystem, the weight values ​​corresponding to the low-sensitivity area, medium-sensitivity area and high-sensitivity area are flexibly introduced in the corresponding calculation process to improve the accuracy of the fire judgment results.

[0062] In this embodiment, the temperature data, carbon monoxide concentration data and ultraviolet light intensity data of the construction area are collected through the wireless sensor network, the sensor nodes are clustered according to the construction area classification results, and the temperature data, carbon monoxide concentration data and ultraviolet light intensity data are sent to the data processing center subsystem through the leader node in the cluster.

[0063] In this embodiment, the wireless sensor network implements leap-layer data transmission by adding an inter-layer interface between the network layer, MAC layer and physical layer in the routing protocol, and periodically conducts leadership node election by calculating the energy coefficient, and the calculation expression is:

[0064] ;

[0065] Where U represents the energy coefficient, Indicates the maximum power of the node. Indicates the remaining power of the candidate node, represents the paradigm distance between the candidate node and other nodes in the cluster, It represents the cumulative sum of the normalized distances from all nodes in the cluster to the communication base station.

[0066] In this embodiment, an environmental data set is established by using the temperature data, carbon monoxide concentration data, and ultraviolet light intensity data collected by the wireless sensor network. The confidence level probability model is obtained by calculating the environmental data set through the DS theory. The confidence level is initialized and assigned to the sets corresponding to the temperature data, carbon monoxide concentration data, and ultraviolet light intensity data. Then, the confidence level is redistributed according to the environmental data newly collected by the wireless sensor network. As the collected data accumulates, the confidence level of each data set is continuously corrected. All data sets are traversed, and probability screening is performed by combining the confidence interval of the sensitivity level and the fire judgment result is output. The calculation expression is:

[0067] ;

[0068] in, represents the trust function that the data set should respond to, represents the likelihood function corresponding to the data set, represents the exact probability, Indicates the sensitivity weight corresponding to the sensor position. In the specific calculation, when the data source area is a low-sensitivity area, the sensitivity weight is used When the data source area is the medium sensitive area, the sensitivity weight is When the data source area is a high-sensitivity area, the sensitivity weight is .

[0069] In this embodiment, the video is sampled and segmented by the frame difference method, the image of the suspicious area is obtained by subtracting adjacent frame images, the flame feature and smoke feature of the suspicious area are extracted, the flame feature vector is obtained by calculating the image information entropy and the brightness change rate of the suspicious area, the smoke feature vector is obtained by calculating the area growth rate and the centroid deviation of the suspicious area, the extracted flame feature vector and smoke feature vector are input into the trained back propagation neural network to obtain the output prediction value, and the fire judgment result is obtained by the fixed threshold detection method, and the calculation expression is:

[0070] ;

[0071] in, represents image information entropy, k represents grayscale value, G represents the maximum grayscale value, represents the probability that the gray value of a pixel is k, ln represents the natural logarithm, represents the predicted value, Indicates the fire existence judgment, Indicates the fixed threshold value of the image. In the specific calculation, when the data source area is a low-sensitivity area, the fixed threshold value is used When the data source area is the medium sensitive area, the fixed threshold value is used When the data source area is a high-sensitivity area, the fixed threshold value is used , obtained When the output result is 1, it is judged that there is a fire in the data source area, and the obtained When the output result is 0, it is judged that there is no fire in the data source area. The training method of the back propagation neural network is as follows: 906 target images are obtained by subtracting adjacent frame images of the video with fire, 3624 fire area images are obtained by horizontally flipping, vertically flipping and horizontally and vertically flipping the target images respectively, flame features and smoke features are extracted from the fire area images, flame feature vectors are obtained by calculating the image information entropy and brightness change rate of the fire area, smoke feature vectors are obtained by calculating the area growth rate and centroid offset of the fire area, the extracted flame feature vectors and smoke feature vectors are used as feature data sets, 80% of the feature data sets are randomly selected as training samples, and the remaining 20% ​​of the feature data sets are used as test samples to complete the training of the back propagation neural network.

[0072] In this embodiment, the fire judgment result is received through the terminal subsystem, a fire alarm is issued at the construction site through the sound and light alarm module, and a notification is sent to the fire department through the cloud platform.

[0073] Example 2, reference Figure 2 , is an embodiment of the present invention, and provides a building construction fire detection system based on the Internet of Things, which specifically includes: a grading subsystem, a wireless sensor network monitoring subsystem, an optical monitoring subsystem, a data processing center subsystem and a terminal subsystem;

[0074] The grading subsystem includes a punch-in module, an area division module and a sensitivity grading module, which are used to divide the construction site into areas and grade the regional sensitivity through manual labeling, and send the construction area classification results and sensitivity grading results to the data processing center subsystem. The area division module and the sensitivity grading module are connected to the data processing center subsystem, and the grading subsystem is connected to the data processing center subsystem;

[0075] The wireless sensor network detection subsystem includes a hybrid sensor network module, a leader node self-checking module and a wireless communication module, which are used to collect environmental data through the wireless sensor network, cluster the sensor nodes according to the construction area classification result, complete the leader node election according to the calculated energy coefficient of the nodes in the cluster, and send the collected environmental data to the data processing center subsystem through the leader node. The wireless sensor network detection subsystem is connected to the data processing center subsystem;

[0076] The optical detection subsystem includes a video acquisition module, a video storage module and a data transmission module, which are used to record a fixed monitoring area to obtain video data and send the video data to the data processing center subsystem in real time; the optical detection subsystem is connected to the data processing center subsystem;

[0077] The data processing center subsystem includes a data aggregation module, a wireless sensor detection data processing module, an optical detection data processing module and a fire judgment module, which are used to establish an environmental data set based on various sensor data collected by the wireless sensor network, calculate a confidence level probability model based on the DS theory and the environmental data set, perform probability screening based on the confidence interval combined with the sensitivity level to obtain a fire judgment result, process the video data based on the frame difference method to obtain a suspicious area image, extract the flame features and smoke features of the suspicious area image, and obtain the fire judgment result through the back propagation neural network and fixed threshold detection method;

[0078] The terminal subsystem includes a cloud platform module, an audible and visual alarm module, and a fire alarm module, which are used to send fire warnings based on the fire judgment results, including sending audible and visual alarms at the construction site and broadcasting fire notifications on the cloud platform, completing fire detection and early warning;

[0079] Wherein, a data set is established according to the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network, the sensor nodes are clustered according to the construction area classification result, and the temperature data, carbon monoxide concentration data and ultraviolet light intensity data are sent to the data processing center subsystem through the leader node in the cluster;

[0080] An inter-layer interface is added between the network layer, MAC layer and physical layer in the routing protocol of the wireless sensor network to realize leap-layer data transmission. The energy coefficients of the nodes in the cluster are calculated regularly, the energy coefficients are sorted, and the node with the largest energy coefficient is selected as the leader node. The calculation expression is:

[0081] ;

[0082] Where U represents the energy coefficient, Indicates the maximum power of the node. Indicates the remaining power of the candidate node, represents the paradigm distance between the candidate node and other nodes in the cluster, Represents the cumulative sum of the paradigm distances from all nodes in the cluster to the communication base station;

[0083] An environmental data set is established based on the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network. The confidence level probability model is calculated for the environmental data set using the DS model. Probability screening is performed based on the confidence interval combined with the sensitivity level and the fire judgment result is output. The calculation expression is:

[0084] ;

[0085] in, represents the trust function corresponding to the environmental dataset, represents the likelihood function corresponding to the data set, represents the exact probability, Indicates the sensitivity weight corresponding to the sensor position;

[0086] The video is sampled and segmented according to the frame difference method. The image of the suspicious area is obtained by subtracting adjacent frame images. The flame feature and smoke feature of the suspicious area are extracted. The image information entropy and brightness change rate of the suspicious area are calculated to obtain the flame feature vector. The area growth rate and centroid offset of the suspicious area are calculated to obtain the smoke feature vector. The extracted flame feature vector and smoke feature vector are input into the trained back propagation neural network. The fire judgment result is obtained by the fixed threshold detection method. The calculation expression is:

[0087] ;

[0088] in, represents image information entropy, k represents grayscale value, G represents the maximum grayscale value, represents the probability that the gray value of a pixel is k, ln represents the natural logarithm, represents the predicted value, Indicates the fire existence judgment, Indicates the fixed threshold value of the image.

[0089] An electronic device comprises a memory, a processor and a computer program stored in the memory and executable on the processor. When the processor executes the computer program, the building construction fire detection system based on the Internet of Things described in Example 2 is executed.

[0090] A non-transitory computer-readable storage medium having a computer program stored thereon, wherein when the computer program is executed by a processor, the building construction fire detection system based on the Internet of Things described in Example 2 is run.

[0091] It should be understood by those skilled in the art that the 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 aspects. Moreover, the present invention may take the form of a computer program product implemented on one or more computer-usable storage media containing computer-usable program codes. Among them, the storage medium may be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as static random access memory (Static Random Access Memory, referred to as SRAM), electrically erasable programmable read-only memory (Electrically Erasable Programmable Read-Only Memory, referred to as EEPROM), erasable programmable read-only memory (Erasable Programmable Read Only Memory, referred to as EPROM), programmable read-only memory (Programmable Red-Only Memory, referred to as PROM), read-only memory (Read-Only Memory, referred to as ROM), magnetic memory, flash memory, magnetic disk or optical disk. 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.

[0092] It should be noted that the above embodiments are only used to illustrate the technical solutions of the present invention rather than to limit it. Although the present invention has been described in detail with reference to the preferred embodiments, those skilled in the art should understand that the technical solutions of the present invention may be modified or replaced by equivalents without departing from the spirit and scope of the technical solutions of the present invention, which should all be included in the scope of the claims of the present invention.

Claims

1. A construction fire detection method based on the Internet of Things, characterized in that: The steps include: Step S1, dividing the construction site into regions and grading the regional sensitivity through manual labeling, and sending the construction area classification results and sensitivity grading results to the data processing center subsystem; Step S2, collecting environmental data through a wireless sensor network, clustering sensor nodes according to the construction area classification result, completing the leader node election according to the calculated energy coefficient of the nodes in the cluster, and sending the collected environmental data to the data processing center subsystem through the leader node; Step S3, recording the fixed monitoring area to obtain video data, and sending the video data to the data processing center subsystem in real time; Step S4, establishing an environmental data set based on various sensor data collected by the wireless sensor network, calculating a confidence level probability model based on DS theory and the environmental data set, and performing probability screening according to the confidence interval combined with the sensitivity level to obtain a fire judgment result; Step S5, processing the video data according to the frame difference method to obtain a suspicious area image, extracting the flame features and smoke features of the suspicious area image, and obtaining a fire judgment result through a back propagation neural network and a fixed threshold detection method; Step S6, sending a fire warning based on the fire judgment result, including sending out an audible and visual alarm at the construction site and broadcasting a fire notification on the cloud platform, completing fire detection and early warning; A data set is established based on the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network, the sensor nodes are clustered according to the construction area classification result, and the temperature data, carbon monoxide concentration data and ultraviolet light intensity data are sent to the data processing center subsystem through the leader node in the cluster; An inter-layer interface is added between the network layer, MAC layer and physical layer in the routing protocol of the wireless sensor network to realize leap-layer data transmission. The energy coefficients of the nodes in the cluster are calculated regularly, the energy coefficients are sorted, and the node with the largest energy coefficient is selected as the leader node. The calculation expression is: ； Where U represents the energy coefficient, Indicates the maximum power of the node. Indicates the remaining power of the candidate node, represents the paradigm distance between the candidate node and other nodes in the cluster, Represents the cumulative sum of the paradigm distances from all nodes in the cluster to the communication base station; An environmental data set is established based on the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network. The confidence level probability model is calculated for the environmental data set using the DS model. Probability screening is performed based on the confidence interval combined with the sensitivity level and the fire judgment result is output. The calculation expression is: ； in, represents the trust function corresponding to the environmental dataset, represents the likelihood function corresponding to the data set, represents the exact probability, Indicates the sensitivity weight corresponding to the sensor position; The video is sampled and segmented according to the frame difference method. The image of the suspicious area is obtained by subtracting adjacent frame images. The flame feature and smoke feature of the suspicious area are extracted. The image information entropy and brightness change rate of the suspicious area are calculated to obtain the flame feature vector. The area growth rate and centroid offset of the suspicious area are calculated to obtain the smoke feature vector. The extracted flame feature vector and smoke feature vector are input into the trained back propagation neural network. The fire judgment result is obtained by the fixed threshold detection method. The calculation expression is: ； in, represents image information entropy, k represents grayscale value, G represents the maximum grayscale value, represents the probability that the gray value of a pixel is k, ln represents the natural logarithm, represents the predicted value, Indicates the fire existence judgment, Indicates the fixed threshold value of the image.

2. The method for fire detection in building construction based on the Internet of Things as claimed in claim 1, characterized in that: Step S11, spatially classifying the construction site, and dividing the construction site into a flammable material area, an open fire operation area, and a general construction area according to the manual marking results of the construction workers; Step S12, spatially classify the construction site, and enter the work mode by having the construction personnel clock in to work, classify the open flame operation area as a low-sensitivity area, the ordinary construction area as a medium-sensitivity area, and the flammable material area as a high-sensitivity area; Construction workers punch in and out to enter off-duty mode, and all construction site areas are classified as high-sensitivity areas.

3. The method for fire detection in building construction based on the Internet of Things as claimed in claim 1, characterized in that: The fire judgment result is received through the terminal subsystem. When the fire judgment result is that a fire exists, an audible and visual alarm is issued at the construction site, and a fire notification is sent to the construction personnel and the fire department through the cloud platform.

4. The building construction fire detection system based on the Internet of Things is characterized by: include: Hierarchical subsystem, wireless sensor network monitoring subsystem, optical monitoring subsystem, data processing center subsystem and terminal subsystem; The grading subsystem includes a punch-in module, an area division module and a sensitivity grading module, which are used to divide the construction site into areas and grade the regional sensitivity through manual labeling, and send the construction area classification results and sensitivity grading results to the data processing center subsystem. The area division module and the sensitivity grading module are connected to the data processing center subsystem, and the grading subsystem is connected to the data processing center subsystem; The wireless sensor network detection subsystem includes a hybrid sensor network module, a leader node self-checking module and a wireless communication module, which are used to collect environmental data through the wireless sensor network, cluster the sensor nodes according to the construction area classification result, complete the leader node election according to the calculated energy coefficient of the nodes in the cluster, and send the collected environmental data to the data processing center subsystem through the leader node. The wireless sensor network detection subsystem is connected to the data processing center subsystem; The optical detection subsystem includes a video acquisition module, a video storage module and a data transmission module, which are used to record a fixed monitoring area to obtain video data and send the video data to the data processing center subsystem in real time; the optical detection subsystem is connected to the data processing center subsystem; The data processing center subsystem includes a data aggregation module, a wireless sensor detection data processing module, an optical detection data processing module and a fire judgment module, which are used to establish an environmental data set based on various sensor data collected by the wireless sensor network, calculate a confidence level probability model based on the DS theory and the environmental data set, perform probability screening based on the confidence interval combined with the sensitivity level to obtain a fire judgment result, process the video data based on the frame difference method to obtain a suspicious area image, extract the flame features and smoke features of the suspicious area image, and obtain the fire judgment result through the back propagation neural network and fixed threshold detection method; The terminal subsystem includes a cloud platform module, an audible and visual alarm module, and a fire alarm module, which are used to send fire warnings based on the fire judgment results, including sending audible and visual alarms at the construction site and broadcasting fire notifications on the cloud platform, completing fire detection and early warning; Wherein, a data set is established according to the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network, the sensor nodes are clustered according to the construction area classification result, and the temperature data, carbon monoxide concentration data and ultraviolet light intensity data are sent to the data processing center subsystem through the leader node in the cluster; An inter-layer interface is added between the network layer, MAC layer and physical layer in the routing protocol of the wireless sensor network to realize leap-layer data transmission. The energy coefficients of the nodes in the cluster are calculated regularly, the energy coefficients are sorted, and the node with the largest energy coefficient is selected as the leader node. The calculation expression is: ； Where U represents the energy coefficient, Indicates the maximum power of the node. Indicates the remaining power of the candidate node, represents the paradigm distance between the candidate node and other nodes in the cluster, Represents the cumulative sum of the paradigm distances from all nodes in the cluster to the communication base station; An environmental data set is established based on the temperature data, carbon monoxide concentration data and ultraviolet light intensity data collected by the wireless sensor network. The confidence level probability model is calculated for the environmental data set using the DS model. Probability screening is performed based on the confidence interval combined with the sensitivity level and the fire judgment result is output. The calculation expression is: ； in, represents the trust function corresponding to the environmental dataset, represents the likelihood function corresponding to the data set, represents the exact probability, Indicates the sensitivity weight corresponding to the sensor position; The video is sampled and segmented according to the frame difference method. The image of the suspicious area is obtained by subtracting adjacent frame images. The flame feature and smoke feature of the suspicious area are extracted. The image information entropy and brightness change rate of the suspicious area are calculated to obtain the flame feature vector. The area growth rate and centroid offset of the suspicious area are calculated to obtain the smoke feature vector. The extracted flame feature vector and smoke feature vector are input into the trained back propagation neural network. The fire judgment result is obtained by the fixed threshold detection method. The calculation expression is: ； in, represents image information entropy, k represents grayscale value, G represents the maximum grayscale value, represents the probability that the gray value of a pixel is k, ln represents the natural logarithm, represents the predicted value, Indicates the fire existence judgment, Indicates the fixed threshold value of the image.

5. An electronic device comprising a memory, a processor, and a computer program stored in the memory and executable on the processor, wherein: When the processor executes the program, the steps of the building construction fire detection method based on the Internet of Things as claimed in any one of claims 1 to 4 are implemented.

6. A non-transitory computer-readable storage medium having a computer program stored thereon, characterized in that: When the computer program is executed by a processor, the steps of the building construction fire detection method based on the Internet of Things as claimed in any one of claims 1 to 4 are implemented.

Images