Factory fire-fighting hidden danger troubleshooting method, device, equipment and medium

Abstract

The invention relates to a factory fire-fighting hidden danger troubleshooting method, device and equipment and a medium, and is applied to the technical field of fire-fighting safety, and the method comprises the steps: obtaining monitoring data and video data; analyzing the monitoring data and the video data to obtain a fire-fighting hidden danger condition; acquiring a historical fire-fighting hidden danger condition; and generating a hidden danger processing suggestion based on the fire-fighting hidden danger condition and the historical fire-fighting hidden danger condition. The method has the effect of improving the fire-fighting hidden danger troubleshooting processing efficiency.

Description

Technical Field

[0001] This application relates to the technical field of fire safety, and in particular to a method, device, equipment and medium for investigating fire hazards in factories. Background Technology

[0002] With the booming development of industry, factories are expanding in scale, production equipment is becoming increasingly complex, and the use of various flammable and explosive materials is becoming more frequent. This presents factories with increasing challenges to fire safety. Effective fire hazard investigation is crucial for factories to prevent fire accidents, protect the lives and property of workers, and maintain normal production order. Once a fire occurs in a factory, if fire hazards are not detected and addressed in a timely manner, it will not only lead to factory shutdown and huge economic losses, but may also trigger a chain reaction, endangering the safety of surrounding areas. Therefore, the research and improvement of factory fire hazard investigation methods have always been of great concern.

[0003] Currently, there are two main methods for identifying fire hazards in factories: one method involves assigning dedicated personnel to conduct regular on-site inspections of various workshops, warehouses, and other areas within the factory. These inspectors rely on their experience and expertise to check the integrity of fire-fighting facilities, the unobstructed nature of evacuation routes, and the presence of electrical leaks. However, the results of this method are easily influenced by the inspectors' subjective biases, making accurate and timely inspections difficult. The other method involves installing simple sensors within the factory, such as smoke and temperature sensors. These sensors trigger alarms when they detect abnormalities. However, simple sensors can only detect specific, single indicators and cannot comprehensively analyze the complex fire hazard situation within the factory, resulting in low efficiency in identifying and addressing fire hazards. Summary of the Invention

[0004] In order to improve the efficiency of fire hazard investigation and handling, this application provides a method, device, equipment and medium for investigating fire hazards in factories.

[0005] Firstly, this application provides a method for investigating fire hazards in factories, employing the following technical solution: A method for identifying fire hazards in factories includes: Acquire monitoring data and video data; The monitoring data and video data are analyzed to determine the fire hazard situation; Obtain historical information on fire hazards; Based on the fire hazard situation and the historical fire hazard situation, hazard handling suggestions are generated.

[0006] By adopting the above technical solutions, the reliability and real-time nature of fire hazard information are improved through comprehensive analysis of monitoring data and video data. When generating hazard handling suggestions, not only the current fire hazard situation is considered, but also the historical fire hazard situation, which improves the reliability of the hazard handling suggestions. By automating the determination of fire hazard situation and handling fire hazard according to the hazard handling suggestions, the efficiency of fire hazard investigation and handling is improved.

[0007] Optionally, the analysis of the monitoring data and the video data to obtain information on fire hazards includes: The monitoring data is then filtered and normalized. The video data is segmented into single-frame image data using a video segmentation tool; The image data is subjected to noise reduction and contrast enhancement processing; The processed monitoring data and image data are analyzed to determine the fire hazard situation.

[0008] By adopting the above technical solution and preprocessing the monitoring and video data, the reliability of the monitoring and video data is improved, providing a reliable basis for subsequent determination of fire hazards.

[0009] Optionally, the step of analyzing the processed monitoring data and image data to determine the fire hazard situation includes: The first feature parameter is obtained by extracting features from the processed monitoring data using the sliding window technique. The processed image data is used to extract features using deep learning technology to obtain the second feature parameters. The first feature parameter and the second feature parameter are fused into a first feature vector; The first feature vector is processed based on a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard information.

[0010] By adopting the above technical solution, feature extraction is performed using the sliding window technique to obtain the first feature parameter, which can then be used to determine the changes and stability of the monitoring data. By extracting and fusing the first feature parameter and the second feature parameter, the comprehensiveness of the features is improved. The first feature vector is processed according to the long short-term memory network model and the multilayer perceptron classification model to obtain the fire hazard situation, thereby improving the reliability of the fire hazard situation.

[0011] Optionally, the step of extracting features from the processed image data using deep learning technology to obtain the second feature parameters includes: The pre-trained ResNet model is loaded based on a deep learning framework; Obtain information on the definition of fire hazards in the factory; The output layer of the ResNet model is adjusted based on the fire hazard definition information. Obtain the training parameters and training dataset; The ResNet model is trained based on the training parameters and the training dataset to obtain the target ResNet model; The second feature parameters are obtained by extracting features from the image data based on the target ResNet model.

[0012] By adopting the above technical solution, the pre-trained ResNet model is fine-tuned using the training dataset corresponding to the factory to obtain the target ResNet model. This allows the target ResNet model to better adapt to the factory scenario while retaining more of the general image features learned during the pre-training process, thereby improving the reliability of the second feature parameters obtained by the target ResNet model.

[0013] Optionally, the process of processing the first feature vector based on a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard information includes: The second feature vector is obtained by performing time-series analysis on the first feature vector using a long short-term memory network model; The second feature vector is processed by a multilayer perceptron classification model to obtain a fire status classification vector; The fire hazard level is determined based on the fire status classification vector; Obtain the SHAP value for each feature dimension in the second feature vector; High-contribution raw data are determined based on the SHAP value and the feature mapping table; The high-contribution raw data is analyzed to identify anomalies; The fire hazard situation is determined based on the abnormal information and the fire hazard level.

[0014] Optionally, generating hazard handling suggestions based on the fire hazard situation and the historical fire hazard situation includes: Based on the aforementioned fire hazard situation, determine the type of anomaly, the location of the anomaly, and the level of the fire hazard; Based on the anomaly type, the anomaly location, and the fire hazard level, the current hazard handling strategy is matched from the strategy database; The fire hazards and historical fire hazards are analyzed to determine hazard prevention strategies; The hazard handling recommendations are determined based on the current hazard handling strategy and the hazard prevention strategy.

[0015] By adopting the above technical solutions, the hazard handling recommendations include both current hazard handling strategies and hazard prevention strategies. This allows staff to not only quickly and accurately handle current fire hazards based on the current hazard handling strategies, but also to prevent future fire hazards based on the hazard prevention strategies.

[0016] Optionally, the analysis of the fire hazard situation and the historical fire hazard situation to determine the hazard prevention strategy includes: Statistical analysis was performed on the fire hazard situation and the historical fire hazard situation to obtain the type abnormality frequency of each of the abnormal types and the location abnormality frequency of each of the abnormal locations; The anomalies whose frequency exceeds the preset anomaly frequency are identified as high-frequency anomalies. An abnormal location whose location anomaly frequency exceeds a preset location anomaly frequency is identified as a high-frequency abnormal location. A first prevention strategy is determined based on the types of high-frequency anomalies. A second prevention strategy is determined based on the high-frequency anomaly location; The hazard prevention strategy is determined based on the first prevention strategy and the second prevention strategy.

[0017] By adopting the above technical solutions and analyzing the current and historical fire hazards, we can obtain the types and locations of high-frequency anomalies, thereby enabling us to determine more targeted hazard prevention strategies.

[0018] Secondly, this application provides a factory fire hazard detection device, which adopts the following technical solution: A factory fire hazard detection device, comprising: The first acquisition module is used to acquire monitoring data and video data; The hazard identification module is used to analyze the monitoring data and the video data to obtain information on fire hazards. The second acquisition module is used to acquire historical fire hazard information; It is recommended that a module be used to generate hazard handling suggestions based on the fire hazard situation and the historical fire hazard situation.

[0019] By adopting the above technical solutions, the reliability and real-time nature of fire hazard information are improved through comprehensive analysis of monitoring data and video data. When generating hazard handling suggestions, not only the current fire hazard situation is considered, but also the historical fire hazard situation, which improves the reliability of the hazard handling suggestions. By automating the determination of fire hazard situation and handling fire hazard according to the hazard handling suggestions, the efficiency of fire hazard investigation and handling is improved.

[0020] Thirdly, this application provides an electronic device that adopts the following technical solution: An electronic device includes a processor coupled to a memory; The memory stores a computer program that can be loaded by a processor and executed as described in any of the first aspects of the factory fire hazard investigation method.

[0021] Fourthly, this application provides a computer-readable storage medium, which adopts the following technical solution: A computer-readable storage medium storing a computer program capable of being loaded by a processor and executing the factory fire hazard investigation method according to any one of the first aspects. Attached Figure Description

[0022] Figure 1 This is a flowchart illustrating a method for investigating fire hazards in a factory, as provided in an embodiment of this application.

[0023] Figure 2 This is a structural block diagram of a factory fire hazard detection device provided in an embodiment of this application.

[0024] Figure 3 This is a structural block diagram of the electronic device provided in the embodiments of this application. Detailed Implementation

[0025] The present application will be further described in detail below with reference to the accompanying drawings.

[0026] This application provides a method for investigating fire hazards in factories. This method can be executed by electronic devices, which can be servers or terminal devices. The server can be a standalone physical server, a server cluster or distributed system composed of multiple physical servers, or a cloud server providing cloud computing services. The terminal device can be a smartphone, tablet, desktop computer, etc., but is not limited to these.

[0027] To make the objectives, technical solutions, and advantages of the embodiments of this application clearer, the technical solutions of the embodiments of this application will be clearly and completely described below with reference to the accompanying drawings. Obviously, the described embodiments are only some embodiments of this application, not all embodiments. Based on the embodiments of this application, all other embodiments obtained by those skilled in the art without creative effort are within the scope of protection of this application.

[0028] Furthermore, the term "and / or" in this article is merely a description of the relationship between related objects, indicating that three relationships can exist. For example, A and / or B can represent: A existing alone, A and B existing simultaneously, or B existing alone. Additionally, the character " / " in this article, unless otherwise specified, generally indicates that the preceding and following related objects have an "or" relationship.

[0029] like Figure 1 As shown, a method for investigating fire hazards in factories is described in the following steps (S101-S104): Step S101: Obtain monitoring data and video data.

[0030] In key areas of the factory, such as production workshops, warehouses, and power distribution rooms, various sensors are deployed, including temperature sensors, humidity sensors, smoke sensors, gas sensors, and pressure sensors. Temperature sensors collect ambient temperature data in real time, humidity sensors collect ambient humidity data in real time, smoke sensors monitor smoke concentration in real time, gas sensors detect gas leaks of flammable, explosive, and toxic gases, and pressure sensors monitor the pressure of fire-fighting pipelines. The monitoring data includes, but is not limited to, temperature data, humidity data, smoke concentration, gas leak data, and fire-fighting pipeline pressure, and is obtained from various sensors.

[0031] Meanwhile, multiple high-definition cameras are installed inside the factory to collect video data from various locations within the factory.

[0032] Step S102: Analyze the monitoring data and video data to obtain information on fire hazards.

[0033] Specifically, the monitoring data and video data are analyzed to determine the fire hazard situation, including: filtering and normalizing the monitoring data; segmenting the video data into single-frame image data using a video segmentation tool; denoising and contrast enhancement of the image data; and analyzing the processed monitoring data and image data to determine the fire hazard situation.

[0034] In this embodiment, the monitoring data and video data are first preprocessed to improve their reliability. Then, the preprocessed monitoring data and video data are used to determine the fire hazard situation, further enhancing the reliability of the fire hazard assessment. The preprocessing specifically includes: filtering the monitoring data using a low-pass filter; normalizing the monitoring data using Python's NumPy library; segmenting the video data into single-frame image data using a video segmentation tool, such as FFmpeg; denoising the image data using a Gaussian denoising algorithm; and enhancing the contrast of the image data using a histogram equalization method. Finally, the preprocessed monitoring data and image data are analyzed to determine the fire hazard situation.

[0035] Furthermore, the processed monitoring data and image data are analyzed to determine the fire hazard situation, including: extracting features from the processed monitoring data using sliding window technology to obtain a first feature parameter; extracting features from the processed image data using deep learning technology to obtain a second feature parameter; fusing the first feature parameter and the second feature parameter into a first feature vector; and processing the first feature vector based on a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard situation.

[0036] In this embodiment, a sliding window technique is used to extract features from the processed monitoring data, resulting in multiple features. These features are then concatenated into a multi-dimensional vector, known as the first feature parameter, in a preset order. Each dimension of the first feature parameter corresponds to a specific feature. The types and number of features for each dimension are pre-defined and not specifically limited here. A 5-minute sliding window can be set, and features such as the mean, variance, and rate of change of the data within each window are calculated. For example, for temperature data, the average temperature, temperature fluctuation variance, and rate of temperature change per minute within a 5-minute sliding window are calculated. These features effectively reflect the temperature change trend and stability.

[0037] The processed image data is used to extract features using deep learning technology to obtain the second feature parameter. The second feature parameter is also represented in the form of a multi-dimensional vector. Each dimension of the second feature parameter corresponds to a feature. The types of features and the number of dimensions corresponding to each dimension of the second feature parameter are preset and are not specifically limited here.

[0038] The first feature parameter and the second feature parameter are fused together to form a first feature vector. For example, if the first feature parameter is (a, b, c) and the second feature parameter is (d, e, f), then the first feature vector is (a, b, c, d, e, f). The first feature vector is then processed by a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard information.

[0039] Specifically, the second feature parameter is obtained by extracting features from the processed image data using deep learning technology, including: loading a pre-trained ResNet model based on a deep learning framework; obtaining fire hazard definition information of the factory; adjusting the output layer of the ResNet model based on the fire hazard definition information; obtaining training parameters and training dataset; training the ResNet model based on the training parameters and training dataset to obtain a target ResNet model; and extracting features from the image data based on the target ResNet model to obtain the second feature parameter.

[0040] In this embodiment, a pre-trained ResNet model is loaded using a deep learning framework (such as TensorFlow, PyTorch, etc.). The pre-trained ResNet model is pre-trained using a large-scale general image dataset (such as ImageNet). Fire hazard definition information for the factory is obtained from a database or from staff. This fire hazard definition information includes object categories, personnel behavior categories, and equipment status categories that need to be identified in the factory fire scenario. The number of categories in the fire hazard definition information is counted, and the number of neurons in the output layer of the ResNet model is set to the number of these categories. Training parameters and a training dataset are obtained from staff or the database. The training parameters are parameters used during model training, such as the network layer freeze range (to better preserve the pre-trained ResNet model during fine-tuning). To train the model to learn general image features and reduce the possibility of overfitting, some layers of the ResNet model can be frozen. Typically, earlier layers learn basic image features such as edges and textures, which are universal across different image tasks. Therefore, freezing these early layers (frozen layers) prevents their parameters from changing during fine-tuning. Later layers, however, can have their parameters updated to learn features specific to the factory scene, such as learning rate, batch size, and number of training epochs. The training dataset consists of labeled images of factory scenes. This training dataset is input into the ResNet model and trained according to the training parameters to obtain the target ResNet model. Image data is then input into the target ResNet model for feature extraction, yielding the second feature parameters.

[0041] Specifically, the first feature vector is processed based on a Long Short-Term Memory (LSTM) network model and a Multilayer Perceptron (MPP) classification model to obtain the fire hazard status. This process includes: performing time-series analysis on the first feature vector using the LTM network model to obtain a second feature vector; processing the second feature vector using the MPP model to obtain a fire status classification vector; determining the fire hazard level based on the fire status classification vector; obtaining the SHAP value for each feature dimension in the second feature vector; identifying high-contribution raw data based on the SHAP values ​​and a feature mapping table; analyzing the high-contribution raw data to identify anomalies; and determining the fire hazard status based on the anomalies and the fire hazard level.

[0042] In this embodiment, the first feature vector is organized into sequential data according to time order and input into a Long Short-Term Memory (LSTM) network model for time-series analysis to obtain a second feature vector. The second feature vector is then input into a Multilayer Perceptron (MPP) classification model to obtain a fire status classification vector. Both the LSM and MPP models are pre-trained and debugged. The vector dimension of the fire status classification vector represents the number of fire hazard levels. Each dimension in the fire status classification vector corresponds to a fire hazard level, and the vector value of a dimension is the probability value of the corresponding fire hazard level. The fire hazard level with the highest probability value is determined as the current fire hazard level. The SHAP value of each feature dimension in the second feature vector and a feature mapping table are obtained from staff or a database. The larger the absolute value of the SHAP value, the greater the contribution of that feature dimension to the model output. For example, staff can input a test dataset into the MPP classification model and use the SHAP library to calculate the SHAP value of each feature dimension. The feature mapping table represents the mapping relationship between each feature dimension in the second feature vector and various monitoring data types or image data. The first preset number (preset, no specific limit here) of feature dimensions with higher absolute values ​​of SHAP values ​​in the second feature vector are identified as high-contribution dimensions. The monitoring data types or image data corresponding to the high-contribution dimensions in the feature mapping table are identified as high-contribution raw data types. The high-contribution raw data (i.e., monitoring data values ​​or image data) corresponding to each high-contribution raw data type are analyzed to obtain abnormal information. For example, if the high-contribution dimensions in the second feature vector are the 3rd to 6th dimensions, and the 3rd to 5th dimensions all correspond to the temperature data at position A, the temperature data is compared with a preset temperature threshold. If the temperature data is higher than the preset temperature threshold, the temperature data is abnormal. The abnormal information includes temperature abnormality (abnormality type), position A (abnormal position), and abnormal temperature data. The 6th dimension corresponds to the image data at position B. The image data is identified using a preset image recognition model (e.g., a convolutional neural network model). If an abnormality is identified, abnormal information is generated based on the identified abnormality. The abnormal information includes the abnormality type, abnormal position, and image data. The abnormal information and the fire hazard level are combined to determine the fire hazard situation.

[0043] Step S103: Obtain historical fire hazard information.

[0044] Historical fire hazard information is retrieved from the database, including the types of historical anomalies, their locations, and their levels.

[0045] Step S104: Generate hazard handling suggestions based on the fire hazard situation and historical fire hazard situation.

[0046] Specifically, based on the current fire hazard situation and historical fire hazard situation, hazard handling suggestions are generated, including: determining the type, location, and level of the anomaly based on the fire hazard situation; matching the current hazard handling strategy from the strategy database based on the anomaly type, location, and level of the fire hazard; analyzing the current fire hazard situation and historical fire hazard situation to determine the hazard prevention strategy; and determining hazard handling suggestions based on the current hazard handling strategy and the hazard prevention strategy.

[0047] In this embodiment, the types, locations, and levels of anomalies are identified from the current fire hazard situation. The strategy database stores the correspondence between anomaly types, locations, fire hazard levels, and hazard handling strategies. The current hazard handling strategy is matched from the strategy database based on the anomaly type, location, and fire hazard level. The current fire hazard situation and historical fire hazard situations are analyzed to determine hazard prevention strategies. The current hazard handling strategy and the hazard prevention strategy are jointly determined as the hazard handling recommendation.

[0048] More specifically, the analysis of current and historical fire hazards determines hazard prevention strategies, including: statistical analysis of current and historical fire hazards to obtain the frequency of each type of anomaly and the frequency of each location anomaly; identifying anomalies whose frequency exceeds a preset frequency as high-frequency anomalies; identifying locations whose frequency exceeds a preset location anomaly as high-frequency anomalies; determining a first prevention strategy based on high-frequency anomaly types; determining a second prevention strategy based on high-frequency anomaly locations; and determining a hazard prevention strategy based on the first and second prevention strategies.

[0049] In this embodiment, data analysis tools (e.g., EXCEL) are used to statistically analyze fire hazard situations and historical fire hazard situations to obtain the frequency of each type of anomaly (e.g., 1 time / month) and the frequency of each location of anomaly. Anomalies whose frequency exceeds a preset frequency are identified as high-frequency anomalies. Anomalies whose frequency exceeds a preset location frequency are identified as high-frequency anomalies. The database stores prevention strategies corresponding to each type of anomaly and prevention strategies corresponding to each location of anomaly. A first prevention strategy is retrieved from the database based on the high-frequency anomaly type, and a second prevention strategy is retrieved from the database based on the high-frequency anomaly location. The first and second prevention strategies are then used together to determine the hazard prevention strategy.

[0050] Figure 2 This is a structural block diagram of a factory fire hazard detection device 200 provided in an embodiment of this application.

[0051] like Figure 2 As shown, the factory fire hazard detection device 200 mainly includes: The first acquisition module 201 is used to acquire monitoring data and video data; The hazard identification module 202 is used to analyze monitoring data and video data to obtain information on fire hazards. The second acquisition module 203 is used to acquire historical fire hazard information; It is recommended to define module 204, which is used to generate hazard handling suggestions based on the current fire hazard situation and historical fire hazard situation.

[0052] As an optional implementation of this embodiment, the hazard determination module 202 is specifically used to analyze monitoring data and video data to obtain fire hazard information, including: filtering and normalizing the monitoring data; segmenting the video data into single-frame image data using a video segmentation tool; denoising and contrast enhancement of the image data; and analyzing the processed monitoring data and image data to determine the fire hazard information.

[0053] As an optional implementation of this embodiment, the hazard determination module 202 is specifically used to analyze the processed monitoring data and image data to determine the fire hazard situation, including: extracting features from the processed monitoring data using sliding window technology to obtain a first feature parameter; extracting features from the processed image data using deep learning technology to obtain a second feature parameter; fusing the first feature parameter and the second feature parameter into a first feature vector; and processing the first feature vector based on a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard situation.

[0054] As an optional implementation of this embodiment, the hazard determination module 202 is specifically used to extract features from the processed image data using deep learning technology to obtain second feature parameters, including: loading a pre-trained ResNet model based on a deep learning framework; obtaining fire hazard definition information of the factory; adjusting the output layer of the ResNet model based on the fire hazard definition information; obtaining training parameters and a training dataset; training the ResNet model based on the training parameters and the training dataset to obtain a target ResNet model; and extracting features from the image data based on the target ResNet model to obtain the second feature parameters.

[0055] As an optional implementation of this embodiment, the hazard determination module 202 is specifically used to process the first feature vector based on a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard situation, including: performing time-series analysis on the first feature vector through the long short-term memory network model to obtain a second feature vector; processing the second feature vector through the multilayer perceptron classification model to obtain a fire status classification vector; determining the fire hazard level based on the fire status classification vector; obtaining the SHAP value of each feature dimension in the second feature vector; determining high-contribution original data based on the SHAP value and the feature mapping table; analyzing the high-contribution original data to determine abnormal information; and determining the fire hazard situation based on the abnormal information and the fire hazard level.

[0056] As an optional implementation of this embodiment, it is suggested that the determining module 204 is specifically used to generate hazard handling suggestions based on the fire hazard situation and historical fire hazard situation, including: determining the type of anomaly, the location of the anomaly, and the level of the fire hazard based on the fire hazard situation; matching the current hazard handling strategy from the strategy database based on the type of anomaly, the location of the anomaly, and the level of the fire hazard; analyzing the fire hazard situation and historical fire hazard situation to determine the hazard prevention strategy; and determining the hazard handling suggestions based on the current hazard handling strategy and the hazard prevention strategy.

[0057] As an optional implementation of this embodiment, it is suggested that the determining module 204 is specifically used to analyze the fire hazard situation and historical fire hazard situation, and determine the hazard prevention strategy, including: performing statistical analysis on the fire hazard situation and historical fire hazard situation to obtain the type abnormality frequency of each type of abnormality and the location abnormality frequency of each abnormal location; determining the abnormality type whose type abnormality frequency exceeds the preset type abnormality frequency as the high-frequency abnormality type; determining the abnormal location whose location abnormality frequency exceeds the preset location abnormality frequency as the high-frequency abnormal location; determining a first prevention strategy based on the high-frequency abnormality type; determining a second prevention strategy based on the high-frequency abnormal location; and determining a hazard prevention strategy based on the first prevention strategy and the second prevention strategy.

[0058] In one example, the module in any of the above devices may be one or more integrated circuits configured to implement the above methods, such as one or more application-specific integrated circuits (ASICs), or one or more digital signal processors (DSPs), or one or more field-programmable gate arrays (FPGAs), or a combination of at least two of these integrated circuit forms.

[0059] For example, when modules in a device can be implemented via a processing element scheduler, the processing element can be a general-purpose processor, such as a central processing unit (CPU) or other processor capable of calling programs. Alternatively, these modules can be integrated together as a system-on-a-chip (SOC).

[0060] Those skilled in the art will clearly understand that, for the sake of convenience and brevity, the specific working process of the above-described device and module can be referred to the corresponding process in the foregoing method embodiments, and will not be repeated here.

[0061] Figure 3 This is a structural block diagram of an electronic device 300 provided in an embodiment of this application.

[0062] like Figure 3 As shown, the electronic device 300 includes a processor 301 and a memory 302, and may further include one or more of an information input / output (I / O) interface 303, a communication component 304, and a communication bus 305.

[0063] The processor 301 controls the overall operation of the electronic device 300 to complete all or part of the steps of the aforementioned factory fire hazard investigation method. The memory 302 stores various types of data to support the operation of the electronic device 300. This data may include, for example, instructions for any application or method operating on the electronic device 300, as well as application-related data. The memory 302 can be implemented by any type of volatile or non-volatile storage device or a combination thereof, such as one or more of Static Random Access Memory (SRAM), Electrically Erasable Programmable Read-Only Memory (EEPROM), Erasable Programmable Read-Only Memory (EPROM), Programmable Read-Only Memory (PROM), Read-Only Memory (ROM), magnetic storage, flash memory, magnetic disk, or optical disk.

[0064] I / O interface 303 provides an interface between processor 301 and other interface modules, such as keyboards, mice, and buttons. These buttons can be virtual or physical. Communication component 304 is used for wired or wireless communication between electronic device 300 and other devices. Wireless communication includes Wi-Fi, Bluetooth, Near Field Communication (NFC), 2G, 3G, or 4G, or a combination thereof. Therefore, the corresponding communication component 304 may include a Wi-Fi component, a Bluetooth component, and an NFC component.

[0065] The electronic device 300 may be implemented by one or more application-specific integrated circuits (ASICs), digital signal processors (DSPs), digital signal processing devices (DSPDs), programmable logic devices (PLDs), field-programmable gate arrays (FPGAs), controllers, microcontrollers, microprocessors, or other electronic components to perform the factory fire hazard investigation method given in the above embodiments.

[0066] The communication bus 305 may include a path for transmitting information between the aforementioned components. The communication bus 305 may be a PCI (Peripheral Component Interconnect) bus or an EISA (Extended Industry Standard Architecture) bus, etc. The communication bus 305 may be divided into an address bus, a data bus, a control bus, etc.

[0067] Electronic device 300 may include, but is not limited to, mobile terminals such as mobile phones, laptops, digital radio receivers, PDAs (personal digital assistants), PADs (tablet computers), PMPs (portable multimedia players), and in-vehicle terminals (such as in-vehicle navigation terminals), as well as fixed terminals such as digital TVs and desktop computers, and may also be servers.

[0068] This application also provides a computer-readable storage medium storing a computer program, which, when executed by a processor, implements the steps of the above-described factory fire hazard investigation method.

[0069] The computer-readable storage medium may include various media capable of storing program code, such as USB flash drives, portable hard drives, read-only memory (ROM), random access memory (RAM), magnetic disks, or optical disks.

[0070] The terms “comprising,” “including,” or any other variations thereof are intended to cover non-exclusive inclusion, such that a process, method, article, or apparatus that comprises a list of elements includes not only those elements but also other elements not expressly listed, or elements inherent to such process, method, article, or apparatus.

[0071] The above description is merely a preferred embodiment of this application and an explanation of the technical principles employed. Those skilled in the art should understand that the scope of this application is not limited to technical solutions formed by specific combinations of the above-described technical features, but should also cover other technical solutions formed by arbitrary combinations of the above-described technical features or their equivalents without departing from the foregoing application concept. For example, technical solutions formed by substituting the above features with (but not limited to) technical features with similar functions claimed in this application.

Claims

1. A method for investigating fire hazards in factories, characterized in that, include: Acquire monitoring data and video data; The monitoring data and video data are analyzed to determine the fire hazard situation; Obtain historical information on fire hazards; Based on the fire hazard situation and the historical fire hazard situation, hazard handling suggestions are generated.

2. The method according to claim 1, characterized in that, The analysis of the monitoring data and the video data to obtain information on fire hazards includes: The monitoring data is then filtered and normalized. The video data is segmented into single-frame image data using a video segmentation tool; The image data is subjected to noise reduction and contrast enhancement processing; The processed monitoring data and image data are analyzed to determine the fire hazard situation.

3. The method according to claim 2, characterized in that, The step of analyzing the processed monitoring data and image data to determine the fire hazard situation includes: The first feature parameter is obtained by extracting features from the processed monitoring data using the sliding window technique. The processed image data is used to extract features using deep learning technology to obtain the second feature parameters. The first feature parameter and the second feature parameter are fused into a first feature vector; The first feature vector is processed based on a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard information.

4. The method according to claim 3, characterized in that, The step of extracting features from the processed image data using deep learning technology to obtain second feature parameters includes: The pre-trained ResNet model is loaded based on a deep learning framework; Obtain information on the definition of fire hazards in the factory; The output layer of the ResNet model is adjusted based on the fire hazard definition information. Obtain the training parameters and training dataset; The ResNet model is trained based on the training parameters and the training dataset to obtain the target ResNet model; The second feature parameters are obtained by extracting features from the image data based on the target ResNet model.

5. The method according to claim 3, characterized in that, The first feature vector is processed using a long short-term memory network model and a multilayer perceptron classification model to obtain the fire hazard information, including: The second feature vector is obtained by performing time-series analysis on the first feature vector using a long short-term memory network model; The second feature vector is processed by a multilayer perceptron classification model to obtain a fire status classification vector; The fire hazard level is determined based on the fire status classification vector; Obtain the SHAP value for each feature dimension in the second feature vector; High-contribution raw data are determined based on the SHAP value and the feature mapping table; The high-contribution raw data is analyzed to identify anomalies; The fire hazard situation is determined based on the abnormal information and the fire hazard level.

6. The method according to claim 1, characterized in that, The generation of hazard handling suggestions based on the fire hazard situation and the historical fire hazard situation includes: Based on the aforementioned fire hazard situation, determine the type of anomaly, its location, and the level of fire hazard. Based on the anomaly type, the anomaly location, and the fire hazard level, the current hazard handling strategy is matched from the strategy database; The fire hazards and historical fire hazards are analyzed to determine hazard prevention strategies; The hazard handling recommendations are determined based on the current hazard handling strategy and the hazard prevention strategy.

7. The method according to claim 6, characterized in that, The analysis of the fire hazard situation and the historical fire hazard situation to determine the hazard prevention strategy includes: Statistical analysis was performed on the fire hazard situation and the historical fire hazard situation to obtain the type abnormality frequency of each of the abnormal types and the location abnormality frequency of each of the abnormal locations; The anomalies whose frequency exceeds the preset anomaly frequency are identified as high-frequency anomalies. An abnormal location whose location anomaly frequency exceeds a preset location anomaly frequency is identified as a high-frequency abnormal location. A first prevention strategy is determined based on the types of high-frequency anomalies. A second prevention strategy is determined based on the high-frequency anomaly location; The hazard prevention strategy is determined based on the first prevention strategy and the second prevention strategy.

8. A factory fire hazard detection device, characterized in that, include: The first acquisition module is used to acquire monitoring data and video data; The hazard identification module is used to analyze the monitoring data and the video data to obtain information on fire hazards. The second acquisition module is used to acquire historical fire hazard information; It is recommended that a module be used to generate hazard handling suggestions based on the fire hazard situation and the historical fire hazard situation.

9. An electronic device, characterized in that, Includes a processor, which is coupled to a memory; The processor is configured to execute a computer program stored in the memory to cause the electronic device to perform the method as described in any one of claims 1 to 7.

10. A computer-readable storage medium, characterized in that, It includes a computer program or instructions that, when run on a computer, cause the computer to perform the method as described in any one of claims 1 to 7.

Images