Substation fire-fighting risk linkage method based on multi-source information fusion and CNN

Abstract

The application discloses a substation fire-fighting risk linkage method based on multi-source information fusion and CNN, standard heterogeneous data is obtained through collection; a fire-fighting risk identification model is constructed based on a convolutional neural network (CNN), flame features and smoke features of image information in the standard heterogeneous data are extracted by using the CNN model, and an image recognition result is output; based on the fusion identification probability, a comprehensive risk assessment function is constructed by combining temperature rise rate and smoke concentration change, a risk value R is calculated; and the corresponding linkage control strategy is dynamically triggered according to the fire state. The fire disaster identification accuracy is significantly improved, and the false alarm rate is reduced.

Description

Technical Field

[0001] This invention relates to the field of substation fire risk management technology, and in particular to a substation fire risk linkage method based on multi-source information fusion and CNN. Background Technology

[0002] As power systems continue to evolve towards digitalization and intelligence, substation operation is gradually shifting from manual to unmanned operation, placing higher demands on fire safety. Traditional fire protection systems mainly rely on single sensors for fire detection, which suffers from limited information sources, insufficient identification accuracy, and delayed response. Existing substation fire protection systems operate independently of each other, lacking an effective collaborative mechanism, making it difficult to meet the rapid response requirements of unmanned scenarios. Summary of the Invention

[0003] The purpose of this invention is to solve the above problems by designing a substation fire risk linkage method based on multi-source information fusion and CNN.

[0004] To achieve the above objectives, the technical solution of the present invention further includes the following steps in the above-mentioned substation fire risk linkage method based on multi-source information fusion and CNN: Real-time acquisition of multi-source data related to fire protection in substations; preprocessing of the multi-source data to obtain standard heterogeneous data; A fire risk identification model is constructed based on a convolutional neural network (CNN). The CNN model is used to extract flame and smoke features from the image information in the standard heterogeneous data and output the image recognition results. The image recognition results are weighted and fused with sensor features to construct a comprehensive discrimination vector, and the fused recognition probability is calculated. The sensor features include smoke concentration, temperature change rate and electrical parameters. Based on the fusion recognition probability, combined with the temperature rise rate and smoke concentration change, a comprehensive risk assessment function is constructed to calculate the risk value R; Based on the risk value R, a classification threshold is set to divide the fire situation into normal state, early warning state, and fire state, and the corresponding linkage control strategy is dynamically triggered according to the fire situation.

[0005] Furthermore, in the above-mentioned substation fire risk linkage method based on multi-source information fusion and CNN, the real-time acquisition of multi-source data related to substation fire protection, and the preprocessing of the multi-source data to obtain standard heterogeneous data, including: Collect multi-source data related to fire protection in substations, including video image information, smoke concentration information, temperature information, and electrical parameter information; The multi-source data is preprocessed, including mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering denoising on the image data, and performing min-max normalization on different modal features to obtain standard heterogeneous data.

[0006] Furthermore, in the aforementioned substation fire risk linkage method based on multi-source information fusion and CNN, the preprocessing of the multi-source data includes mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering denoising on the image data, and performing minimum-maximum normalization standardization on different modal features to obtain standard heterogeneous data, including: Let the multi-source data collected by the system at time t be represented as follows: These correspond to image information, smoke concentration, temperature value, and electrical characteristic parameters, respectively. Time offset correction Various types of data are mapped to a unified time axis, and low-frequency data are reconstructed using linear interpolation methods; An adaptive contrast enhancement method is used to dynamically stretch the image grayscale to highlight the flame and smoke areas, and then grayscale normalization is applied. Improve contrast.

[0007] Furthermore, in the aforementioned substation fire risk linkage method based on multi-source information fusion and CNN, the preprocessing of the multi-source data includes mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering denoising on the image data, and performing minimum-maximum normalization standardization on different modal features to obtain standard heterogeneous data, including: Different modal features are standardized by using a min-max normalization method to map the original features to a uniform scale interval. ; Image depth features, temperature change rate, and smoke concentration are fused together for representation, with the temperature change rate calculated by the difference between adjacent time steps. ; Introducing a weighted fusion strategy to construct multimodal feature vectors: (2); in Representing the modal characteristics, For the corresponding weight and satisfying By adjusting the weights, the expressive power of key features in the model can be enhanced.

[0008] Furthermore, in the above-mentioned substation fire risk linkage method based on multi-source information fusion and CNN, the fire risk identification model constructed based on convolutional neural network (CNN) is used to extract flame and smoke features from the image information in the standard heterogeneous data, and the image recognition result is output, including: The model introduces multi-scale convolutional kernels on the basic convolutional structure to enhance the ability to perceive the features of flames and smoke of different sizes. At the same time, it combines a feature pyramid structure to achieve the fusion of high-level semantic information and low-level detailed information. The basic operation process of a convolutional layer is represented as follows: (3); in, Let (l) be the feature map of layer (l), and (*) denote the convolution operation. and These are weights and biases, respectively. It is a non-linear activation function; The CNN output is fused with sensor data to make a decision. Let the image recognition output be... The sensor features are A comprehensive discrimination vector is constructed through a fusion layer.

[0009] Furthermore, in the above-mentioned substation fire risk linkage method based on multi-source information fusion and CNN, the image recognition result and sensor features are weighted and fused to construct a comprehensive discrimination vector, and the fused recognition probability is calculated. The sensor features include smoke concentration, temperature change rate, and electrical parameters, including: Based on fusion output probability Construct a fire risk discrimination function and classify fire levels by setting grading thresholds; when Determined to be in a normal state, when The status is determined to be an early warning state. The situation has been determined to be a fire.

[0010] Furthermore, in the above-mentioned substation fire risk linkage method based on multi-source information fusion and CNN, the step of constructing a comprehensive risk assessment function based on the fused identification probability, combined with the temperature rise rate and smoke concentration change, and calculating the risk value R includes: A fire risk classification model is constructed based on the fusion identification probability, fusion identification confidence level, temperature rise rate, and smoke concentration change. Let the fusion identification probability be... The rate of temperature change Let the smoke concentration be (S), then define the comprehensive risk assessment function: (5); in, Assign weights to each indicator, satisfying ; Fire status is classified based on risk value (R) threshold settings: when This is the normal state. In a state of alert, The situation has been determined to be a fire.

[0011] Furthermore, in the system for realizing the substation fire risk linkage method based on multi-source information fusion and CNN, the system includes the following modules; The data acquisition module is used to collect multi-source data related to fire protection in substations in real time, and to preprocess the multi-source data to obtain standard heterogeneous data. The image recognition module is used to build a fire risk recognition model based on a convolutional neural network (CNN), and to use the CNN model to extract flame and smoke features from the image information in the standard heterogeneous data, and output the image recognition results. The weighted fusion module is used to weight and fuse the image recognition results with the sensor features to construct a comprehensive discrimination vector and calculate the fusion recognition probability. The sensor features include smoke concentration, temperature change rate and electrical parameters. The risk value calculation module is used to construct a comprehensive risk assessment function based on the fusion recognition probability, combined with the temperature rise rate and smoke concentration change, and calculate the risk value R. The linkage control module is used to set a classification threshold based on the risk value R, divide the fire status into normal status, early warning status and fire status, and dynamically trigger the corresponding linkage control strategy according to the fire status.

[0012] Furthermore, in the system for implementing a substation fire risk linkage method based on multi-source information fusion and CNN, the data acquisition module includes the following sub-modules: The data acquisition submodule is used to acquire multi-source data related to fire protection in the substation. The multi-source data includes video image information, smoke concentration information, temperature information, and electrical parameter information. The processing submodule is used to preprocess the multi-source data, including mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering noise reduction on the image data, and performing minimum-maximum normalization on different modal features to obtain standard heterogeneous data.

[0013] Furthermore, in the substation fire risk linkage system based on multi-source information fusion and CNN, the risk value calculation module includes the following sub-modules: The calculation submodule is used to construct a fire risk classification model based on the fusion recognition confidence level, temperature rise rate, and smoke concentration change, according to the fusion recognition probability. Let the fusion recognition probability be... The rate of temperature change Let the smoke concentration be (S), then define the comprehensive risk assessment function: (5); in, Assign weights to each indicator, satisfying ; The classification submodule is used to classify fire status based on risk values ​​(R) and setting grading thresholds: when This is the normal state. In a state of alert, The situation has been determined to be a fire.

[0014] Its beneficial effects are as follows: (1) It significantly improves the accuracy of fire identification and reduces the false alarm rate. This invention integrates multi-source heterogeneous information such as video images, smoke concentration, temperature changes and electrical parameters, and combines the deep feature extraction capability of CNN with a weighted fusion collaborative discrimination strategy to effectively distinguish between fire and non-fire features. Experimental results show that the fire identification accuracy reaches 96.5% and the false alarm rate is reduced to 3.8%; compared with the traditional fusion method, the false alarm rate is reduced by about 58.7%, which significantly reduces the risk of system misjudgment and improves operational reliability. (2) It enhances the early fire warning capability. The comprehensive risk assessment function constructed by this invention integrates multi-dimensional indicators such as identification confidence, temperature rise rate and smoke concentration changes. It still maintains an identification accuracy of 93.1% in the early stage of fire (weak fire / smoke), which can effectively capture the weak signals of early fire and buy valuable time for timely intervention. Attached Figure Description

[0015] Various other advantages and benefits will become apparent to those skilled in the art upon reading the following detailed description of preferred embodiments. The accompanying drawings are for illustrative purposes only and are not intended to limit the invention.

[0016] Figure 1 This is a schematic diagram of the first embodiment of the substation fire risk linkage method based on multi-source information fusion and CNN in this invention. Figure 2 This is a schematic diagram of the second embodiment of the substation fire risk linkage method based on multi-source information fusion and CNN in this invention. Detailed Implementation

[0017] To make the objectives, technical solutions, and advantages of this invention clearer, the invention will be further described in detail below with reference to the accompanying drawings and embodiments. It should be understood that the specific embodiments described herein are merely illustrative and not intended to limit the invention.

[0018] Those skilled in the art will understand that, unless specifically stated otherwise, the singular forms "one," "an," and "this" used herein may also include the plural forms. It should be further understood that the terminology used in this specification includes the presence of features, integers, steps, operations, elements, and / or components, but does not preclude the presence or addition of one or more other features, integers, steps, operations, elements, components, and / or groups thereof.

[0019] The present invention will now be described in detail with reference to the accompanying drawings. Figure 1 As shown, the substation fire risk linkage method based on multi-source information fusion and CNN includes the following steps: Step 101: Collect multi-source data related to fire protection in the substation in real time, and preprocess the multi-source data to obtain standard heterogeneous data; Specifically, in this embodiment, multi-source data related to substation fire protection are collected, including video image information, smoke concentration information, temperature information, and electrical parameter information. Let the multi-source data collected by the system at time (t) be represented as follows: These correspond to image information, smoke concentration, temperature value, and electrical characteristic parameters, respectively.

[0020] Time offset correction Various data types are mapped to a unified time axis, and low-frequency data are reconstructed using linear interpolation methods to meet the unified sampling requirements. For example, data at any given time can be interpolated from adjacent sampling points to achieve continuous-time representation.

[0021] An adaptive contrast enhancement method is used to dynamically stretch the image grayscale, making the flame and smoke areas more prominent. Essentially, this is achieved through grayscale normalization. Improve contrast.

[0022] Based on this, Gaussian filtering is introduced to smooth the image and suppress environmental noise interference. Its core process can be represented as follows: (1); in The kernel function is Gaussian. Furthermore, to enhance the characteristics of flame edges and smoke diffusion, a gradient operator is introduced to extract image edge information, that is, to characterize the target contour through the rate of grayscale change.

[0023] To achieve unified modeling of multi-source data, features of different modalities are standardized. A min-max normalization method is used to map the original features to a unified scale interval. This is to eliminate the effects of dimensional differences.

[0024] In terms of feature construction, image depth features, temperature change rate, and smoke concentration are fused together for representation. The temperature change rate is calculated by differencing adjacent time points. It can effectively reflect the development trend of a fire.

[0025] To integrate the contributions of different information sources, a weighted fusion strategy is introduced to construct multimodal feature vectors: (2); in Representing the modal characteristics, For the corresponding weight and satisfying By adjusting the weights, the expressive power of key features in the model can be enhanced.

[0026] Step 102: Construct a fire risk identification model based on a convolutional neural network (CNN), and use the CNN model to extract flame and smoke features from the image information in the standard heterogeneous data, and output the image recognition results; Specifically, in this embodiment, the model introduces multi-scale convolutional kernels (such as...) on top of the basic convolutional structure. , This enhances the ability to perceive the characteristics of flames and smoke of different sizes. At the same time, it combines a feature pyramid structure to achieve the fusion of high-level semantic information and low-level detailed information, thereby improving the robustness of recognition in complex scenarios.

[0027] The basic operation process of a convolutional layer can be represented as follows: (3); in, Let (l) be the feature map of layer (l), and (*) denote the convolution operation. and These are weights and biases, respectively. It is a non-linear activation function.

[0028] After image feature extraction, the CNN output is fused with sensor data for decision-making. Let the image recognition output be... The sensor features are Then, a comprehensive discrimination vector is constructed through the fusion layer.

[0029] To reduce the risk of misjudgment caused by a single information source, a weighted fusion strategy is introduced to collaboratively make decisions based on information from different sources. This can be expressed as: (4); in, The risk probability is calculated based on sensor data. These are the weighting coefficients.

[0030] Based on fusion output probability A fire risk discrimination function is constructed, and fire severity levels are classified by setting grading thresholds. For example, when... Determined to be in a normal state, when The status is determined to be an early warning state. The situation has been determined to be a fire.

[0031] By combining abnormal thresholds from sensors such as temperature and smoke, multiple conditions can be jointly assessed, thereby improving the accuracy of risk identification.

[0032] In terms of spatial positioning, a mapping relationship is established using video surveillance area division and sensor deployment information. For image data, the bounding box position ((x,y)) of the flame area is obtained through target detection algorithm, and mapped to actual spatial coordinates in combination with camera calibration parameters; for sensor data, auxiliary positioning is performed based on the location of abnormal nodes.

[0033] Step 103: The image recognition result and sensor features are weighted and fused to construct a comprehensive discrimination vector and calculate the fusion recognition probability. The sensor features include smoke concentration, temperature change rate and electrical parameters. Step 104: Based on the fusion recognition probability, combined with the temperature rise rate and smoke concentration change, construct a comprehensive risk assessment function and calculate the risk value R; Specifically, in this embodiment, a fire risk classification model is constructed based on the multi-source fusion output results, considering fusion identification confidence, temperature rise rate, and changes in smoke concentration. Let the fusion identification probability be... The rate of temperature change Let the smoke concentration be (S), then define the comprehensive risk assessment function: (5); in, Assign weights to each indicator, satisfying .

[0034] Fire status is classified by setting a grading threshold based on risk value (R): when This is the normal state. In a state of alert, The fire has been identified as a fire. By integrating multi-dimensional information, this model can effectively depict the fire's development process and improve the accuracy and stability of fire classification.

[0035] The system dynamically triggers different control strategies based on the real-time risk level: in the early warning stage, it activates audible and visual alarms and sends alert information; in the fire stage, it links the ventilation system and automatic fire extinguishing devices to achieve rapid intervention.

[0036] To improve the rationality of control decisions, a logic control function is introduced to describe the linkage relationship, that is, the system control output (U) can be expressed as... ,in This is a piecewise function, with different execution strategies corresponding to different risk ranges. This approach establishes a mapping between control actions and risk levels, ensuring the system's response is targeted and effective.

[0037] At the same time, the control strategy is dynamically modified based on the equipment operating status and environmental conditions, so that the system has a certain degree of self-adaptation capability and improves the overall linkage efficiency.

[0038] Step 105: Set a classification threshold based on the risk value R, divide the fire situation into normal state, early warning state and fire state, and dynamically trigger the corresponding linkage control strategy according to the fire situation.

[0039] Specifically, in this embodiment, to adapt to the operational needs of unattended substations, an event-triggered emergency response scheduling mechanism is constructed. When the system detects an abnormal event, it automatically initiates the response process based on the risk level, including alarm push, remote confirmation, and handling execution.

[0040] Its beneficial effects are as follows: (1) It significantly improves the accuracy of fire identification and reduces the false alarm rate. This invention integrates multi-source heterogeneous information such as video images, smoke concentration, temperature changes and electrical parameters, and combines the deep feature extraction capability of CNN with a weighted fusion collaborative discrimination strategy to effectively distinguish between fire and non-fire features. Experimental results show that the fire identification accuracy reaches 96.5% and the false alarm rate is reduced to 3.8%; compared with the traditional fusion method, the false alarm rate is reduced by about 58.7%, which significantly reduces the risk of system misjudgment and improves operational reliability. (2) It enhances the early fire warning capability. The comprehensive risk assessment function constructed by this invention integrates multi-dimensional indicators such as identification confidence, temperature rise rate and smoke concentration changes. It still maintains an identification accuracy of 93.1% in the early stage of fire (weak fire / smoke), which can effectively capture the weak signals of early fire and buy valuable time for timely intervention.

[0041] Please see Figure 2 In the substation fire risk linkage method based on multi-source information fusion and CNN, multi-source data related to substation fire protection are collected in real time, and the multi-source data is preprocessed to obtain standard heterogeneous data, including the following steps. Step 201: Collect multi-source data related to fire protection in the substation. The multi-source data includes video image information, smoke concentration information, temperature information, and electrical parameter information. Step 202: Preprocess the multi-source data, including mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering noise reduction on the image data, and performing minimum-maximum normalization on different modal features to obtain standard heterogeneous data.

[0042] The foregoing has shown and described the basic principles, main features, and advantages of the present invention. Those skilled in the art should understand that the present invention is not limited to the above embodiments. The embodiments and descriptions in the specification are merely preferred examples and are not intended to limit the invention. Various changes and modifications can be made to the invention without departing from its spirit and scope, and all such changes and modifications fall within the scope of the present invention as claimed. The scope of protection of the present invention is defined by the appended technical solutions and equivalents.

Claims

1. A substation fire risk linkage method based on multi-source information fusion and CNN, characterized in that, The substation fire risk linkage method includes the following steps: Real-time acquisition of multi-source data related to fire protection in substations; preprocessing of the multi-source data to obtain standard heterogeneous data; A fire risk identification model is constructed based on a convolutional neural network (CNN). The CNN model is used to extract flame and smoke features from the image information in the standard heterogeneous data and output the image recognition results. The image recognition results are weighted and fused with sensor features to construct a comprehensive discrimination vector, and the fused recognition probability is calculated. The sensor features include smoke concentration, temperature change rate and electrical parameters. Based on the fusion recognition probability, combined with the temperature rise rate and smoke concentration change, a comprehensive risk assessment function is constructed to calculate the risk value R; Based on the risk value R, a classification threshold is set to divide the fire situation into normal state, early warning state, and fire state, and the corresponding linkage control strategy is dynamically triggered according to the fire situation.

2. The substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 1, characterized in that, The real-time acquisition of multi-source data related to substation fire protection involves preprocessing the multi-source data to obtain standard heterogeneous data, including: Collect multi-source data related to fire protection in substations, including video image information, smoke concentration information, temperature information, and electrical parameter information; The multi-source data is preprocessed, including mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering denoising on the image data, and performing min-max normalization on different modal features to obtain standard heterogeneous data.

3. The substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 2, characterized in that, The preprocessing of the multi-source data includes mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering denoising on the image data, and performing minimum-maximum normalization on different modal features to obtain standard heterogeneous data, including: Let the multi-source data collected by the system at time t be represented as follows: These correspond to image information, smoke concentration, temperature value, and electrical characteristic parameters, respectively. Time offset correction Various types of data are mapped to a unified time axis, and low-frequency data are reconstructed using linear interpolation methods; An adaptive contrast enhancement method is used to dynamically stretch the image grayscale to highlight the flame and smoke areas, and then grayscale normalization is applied. Improve contrast.

4. The substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 2, characterized in that, The preprocessing of the multi-source data includes mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering denoising on the image data, and performing minimum-maximum normalization on different modal features to obtain standard heterogeneous data, including: Different modal features are standardized by using a min-max normalization method to map the original features to a uniform scale interval. ; Image depth features, temperature change rate, and smoke concentration are fused together for representation, with the temperature change rate calculated by the difference between adjacent time steps. .

5. The substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 1, characterized in that, The fire risk identification model is constructed based on a convolutional neural network (CNN). This CNN model is used to extract flame and smoke features from image information in the standard heterogeneous data. The output image recognition results include: The model introduces multi-scale convolutional kernels on the basic convolutional structure to enhance the ability to perceive the features of flames and smoke of different sizes. At the same time, it combines a feature pyramid structure to achieve the fusion of high-level semantic information and low-level detailed information.

6. The substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 1, characterized in that, The image recognition result is weighted and fused with sensor features to construct a comprehensive discrimination vector, and the fused recognition probability is calculated. The sensor features include smoke concentration, temperature change rate, and electrical parameters, including: Based on fusion output probability Construct a fire risk discrimination function and classify fire levels by setting grading thresholds; when Determined to be in a normal state, when The status is determined to be an early warning state. The situation has been determined to be a fire.

7. The substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 1, characterized in that, The comprehensive risk assessment function is constructed based on the fused recognition probability, combined with the temperature rise rate and smoke concentration change, and the risk value R is calculated, including: A fire risk classification model is constructed based on the fusion identification probability, fusion identification confidence level, temperature rise rate, and smoke concentration change. Let the fusion identification probability be... The rate of temperature change Let the smoke concentration be (S), then define the comprehensive risk assessment function: ； in, Assign weights to each indicator, satisfying ; Fire status is classified based on risk value (R) threshold settings: when This is the normal state. In a state of alert, The situation has been determined to be a fire.

8. A system for implementing the substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 1, characterized in that, The system includes the following modules: The data acquisition module is used to collect multi-source data related to fire protection in substations in real time, and to preprocess the multi-source data to obtain standard heterogeneous data. The image recognition module is used to build a fire risk recognition model based on a convolutional neural network (CNN), and to use the CNN model to extract flame and smoke features from the image information in the standard heterogeneous data, and output the image recognition results. The weighted fusion module is used to weight and fuse the image recognition results with the sensor features to construct a comprehensive discrimination vector and calculate the fusion recognition probability. The sensor features include smoke concentration, temperature change rate and electrical parameters. The risk value calculation module is used to construct a comprehensive risk assessment function based on the fusion recognition probability, combined with the temperature rise rate and smoke concentration change, and calculate the risk value R. The linkage control module is used to set a classification threshold based on the risk value R, divide the fire status into normal status, early warning status and fire status, and dynamically trigger the corresponding linkage control strategy according to the fire status.

9. A system for implementing the substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 8, characterized in that, The data acquisition module includes the following sub-modules: The data acquisition submodule is used to acquire multi-source data related to fire protection in the substation. The multi-source data includes video image information, smoke concentration information, temperature information, and electrical parameter information. The processing submodule is used to preprocess the multi-source data, including mapping the data to a unified time axis for time synchronization, performing adaptive contrast enhancement and Gaussian filtering noise reduction on the image data, and performing minimum-maximum normalization on different modal features to obtain standard heterogeneous data.

10. A system for implementing the substation fire risk linkage method based on multi-source information fusion and CNN as described in claim 8, characterized in that, The risk value calculation module includes the following sub-modules: The calculation submodule is used to construct a fire risk classification model based on the fusion recognition confidence level, temperature rise rate, and smoke concentration change, according to the fusion recognition probability. Let the fusion recognition probability be... The rate of temperature change Let the smoke concentration be (S), then define the comprehensive risk assessment function: ； in, Assign weights to each indicator, satisfying ; The classification submodule is used to classify fire status based on risk values ​​(R) and setting grading thresholds: when This is the normal state. In a state of alert, The situation has been determined to be a fire.

Images